This is the last place we used the filepath arguments without first using the
context version.
Differential Revision: https://phab.mercurial-scm.org/D379
This will be used as an abstraction by simplemerge to get the data it used to
read off the filesystem.
Differential Revision: https://phab.mercurial-scm.org/D434
This moves the bundle2 part handling for bundlerepo out to a separate function
so extensions can participate in bundlerepo setup when using bundle2 bundles.
Differential Revision: https://phab.mercurial-scm.org/D290
Previously, iterparts would yield the part to users, then consume the part. This
changed the part after the user was given it and left it at the end, both of
which seem unexpected. Let's seek back to the beginning after we've consumed
it. I tried not seeking to the end at all, but that seems important for the
overall bundle2 consumption.
This is used in a future patch to let us move the bundlerepo
bundle2-changegroup-part to be handled entirely within the for loop, instead of
having to do a seek back to 0 after the entire loop finishes.
Differential Revision: https://phab.mercurial-scm.org/D289
A future patch will refactor certain parts of bundlerepo initiatlization such
that we need to create temp bundles from another function. Let's move this to
another function to support that.
Differential Revision: https://phab.mercurial-scm.org/D288
The Mercurial core server doesn't yet include phase-heads parts in the
bundle, but our Google-internal server wants to do
that. Unfortunately, the usual exchange still happens even if
phase-heads part is included (including the short-circuited one for
old/publishing servers). That means that even if our server (again,
the Google-internal one, but also future Mercurial core servers)
includes a phase-heads part to indicate that some heads should be
drafts, that would still get overwritten by the phase updating that
happens after. So let's fix that by marking the phase step done if we
receive at least one phase-heads part in the bundle.
Differential Revision: https://phab.mercurial-scm.org/D440
Setting `repo._shellvars` works but is not a clean way to pass the pushvars
information from the push command to the exchange operation. Therefore
change it to actually pass `pushvars` as a push operation argument instead.
This makes third party extension like remotenames easier to support pushvars
cleanly. The key value parsing and verification code has been moved to a
lower level so it's harder to be bypassed and easier to be used in
remotenames which could replace `push` command entirely.
Differential Revision: https://phab.mercurial-scm.org/D423
There are cases like bisect when the conflict message can be None. So we make
sure that we don't print None in that case.
Thanks to Martin for catching this.
Differential Revision: https://phab.mercurial-scm.org/D461
This enables the optimization introduced by b8d938230143 for non-rebase cases.
Before, the match couldn't be narrowed if it was e.g. alwaysmatcher.
The logic is copied from fca0d99edf8e.
In chg's case, making modules lazily loaded could actually slow down things
since chg pre-imports them. Therefore disable demandimport if chg is being
used.
This is not done by setting `HGDEMANDIMPORT` chg client-side because that
has side-effects on child processes (hooks, etc).
Differential Revision: https://phab.mercurial-scm.org/D351
Wrapping text in templates for 'hg log --graph' can't be done very well,
because the template doesn't know how wide the graph drawing is. The edge
drawing function needs to know the number of lines in the template output, so
we need to also determine how wide that drawing would be before we call the
edgefn or evaluate the template.
This patch makes edgefn compute the graph width and pass it into the template
so that we can do something like this:
COLUMNS=10 hg log --graph --template "{fill(desc, termwidth - graphwidth)}"
@ a a a a
| a a a a
| a a a a
o a a a
|\ a a a
| | a a a
| | a a a
Using extensions to do this would be relatively complicated due to a lack of
hooks in this area of the code.
In the future it may make sense to have a more generic "textwidth" that tells
you how many columns you can expect to fill without causing the terminal to
wrap your output. I'm not sure there are other situations to motivate this yet,
or if it is entirely feasible.
Differential Revision: https://phab.mercurial-scm.org/D360
Update all calls to formatter.write first arguments to remove references to
precnode and use prednode consistently everywhere.
Differential Revision: https://phab.mercurial-scm.org/D414
Seems the code block misses `::`. This patch makes sure `[push]` and
`pushvars.server = true` are in two lines.
Differential Revision: https://phab.mercurial-scm.org/D411
In order to make this work, we have to wrap the io streams in a
TextIOWrapper so that __builtins__.input() can do unicode IO on Python
3. We can't just restore the original (unicode) sys.std* because we
might be running a cmdserver, and if we blindly restore sys.* to the
original values then we end up breaking the cmdserver. Sadly,
TextIOWrapper tries to close the underlying stream during its __del__,
so we have to make a sublcass to prevent that.
If you see errors like:
TypeError: a bytes-like object is required, not 'str'
On an input() or print() call on Python 3, the substitution of
sys.std* is probably the root cause.
A previous version of this change tried to put the bytesinput() method
in pycompat - it turns out we need to do some encoding handling, so we
have to be in a higher layer that's allowed to use
mercurial.encoding.encoding. As a result, this is in util for now,
with the TextIOWrapper subclass hiding in encoding.py. I'm not sure of
a better place for the time being.
Differential Revision: https://phab.mercurial-scm.org/D299
Rename the existing parameters for clarity.
These will, in subsequent patches, allow callers to redirect reads (of the
three sides of the merge) and writes (of the result) to the given contexts,
instead of using the filesystem.
While in most cases, the writes will go to a workingfilectx, this opens the
door for it to be a memfilectx in the case of an in-memory merge.
Repo will be necessary in a subsequent comit.
Differential Revision: https://phab.mercurial-scm.org/D372
When I wrote f33f2b5b4874 (commit: don't let failed commit with
--addremove update dirstate (issue5645), 2017-07-31), Durham's
422ca3501516 (rebase: use one dirstateguard for when using
rebase.singletransaction, 2017-07-20) had not yet landed, so I had to
write it in the old-fashioned way. Now that Durham's patch is in, we
can simplify by using a context manager.
Differential Revision: https://phab.mercurial-scm.org/D406
This is just a simple refactoring to make the next patch simpler. If
the dirstateguard constructor raises an exception, the finally-block
won't do anything anyway, so this is functionally equivalent (and
there is no except-block).
Differential Revision: https://phab.mercurial-scm.org/D405
This documentation is very helpful for any developer to understand what
copytracing is and what the function does. Since this is the main function of
doing copytracing, I have also included bits about copytracing in it.
This additions are picked from a doc by Stash@Fb. So thanks to him.
Differential Revision: https://phab.mercurial-scm.org/D409
This is probably only used in the 'unbundle' command, but the code
ended up being cleaner to make it generic and treat *all* httppostargs
with a non-args request body as though they were file-like in
nature. It also means we get test coverage more or less for free. A
previous version of this change didn't use io.BytesIO, and it was a
lot more complicated.
This also fixes a server-side bug, so anyone using httppostargs should
update all of their servers to this revision or later *before* this
gets to their clients, otherwise servers will hang trying to over-read
the POST body.
Differential Revision: https://phab.mercurial-scm.org/D231
To list up available compression types instead of
".. bundlecompressionmarker" in "hg help bundlespec" output, proxy
object "docobject" is used, because:
- current online help system requires that __doc__ of registered
object (maybe, function) is already well formatted in reST syntax
- bundletype() method of compressionengine classes is used to list up
available compression types, but
- __doc__ of bundletype() object (= "instancemethod") is read-only
On the other hand, hggettext requires original function object, in
order to get document location in source code.
Therefore, description of each compression types isn't yet
translatable. Even if translatable, translators should make much
effort to determine location of original texts in source code.
To get actual function information, this patch makes hggettext use
function object saved as "_origfunc", if it is available. This patch
also changes bundlecompressiontopics() side, in order to explain how
these changes work easily.
This patch is a part of preparations for making description of each
compression types translatable.
The status line in the crecord has the "space" status field which has variable
length depending on the length of the status label in the language of choice.
In English, the status labels are "space: deselect" and "space:select". The
"deselect" label is 2 glyphs longer. This makes the terminal output jump
around if the terminal width is just right so that the shorter label makes
the status line 1 line long, and the longer label makes it 2 lines long.
This patch formats the selected status into a fixed-width field. The field
width is the maximum of the lengths of the two possible labels, to account for
differing translations and label lengths. This should make the label behavior
uniform across localizations.
There does not seem to be a test for crecord, so I verified the change manually
with a local build of 'hg'.
The wirepeer class provides concrete implementations of peer interface
methods for calling wire protocol commands. It makes sense for this
class to inherit from the peer abstract base class. So we change
that.
Since httppeer and sshpeer have already been converted to the new
interface, peerrepository is no longer adding any value. So it has
been removed. httppeer and sshpeer have been updated to reflect the
loss of peerrepository and the inheritance of the abstract base
class in wirepeer.
The code changes in wirepeer are reordering of methods to group
by interface.
Some Python code in tests was updated to reflect changed APIs.
.. api::
peer.peerrepository has been removed. Use repository.peer abstract
base class to represent a peer repository.
Differential Revision: https://phab.mercurial-scm.org/D338
We need the same @property conversion of ui like we did for localpeer.
We renamed _capabilities() to capabilities() to satisfy the new
naming requirement.
However, since we're inheriting from wireproto.wirepeer which inherits
from peer.peerrepository and provides its own code accessing
_capabilities(), we need to keep the old alias around. This wonkiness
will disappear once wirepeer is cleaned up in subsequent commits.
We also implement methods for basepeer that are identical to the
defaults in peer.peerrepository in preparation for the removal of
peerrepository.
Differential Revision: https://phab.mercurial-scm.org/D336
We now have a formal abstract base class for peers. Let's
transition the peer classes in localrepo to it.
As part of the transition, we reorder methods so they are grouped
by interface and match the order they are defined in the interface.
We also had to change self.ui from an instance attribute to a
property to satisfy the @abstractproperty requirement.
As part of this change, we uncover the first "bug" as part of
enforcing interfaces: stream_out() wasn't implemented on localpeer!
This isn't technically a bug since the repo isn't advertising the
stream capability, so clients shouldn't be attempting to call it.
But I don't think there's a good reason why this is the case.
We implement a dummy method to satisfy the interface requriements.
We can make localpeer instances streamable as a future enhancement.
# no-check-commit
Differential Revision: https://phab.mercurial-scm.org/D335
These methods are part of the peer interface, are generic, and can
be implemented in terms of other members of the peer interface. So we
implement them on the peer base class as a convenience.
The implementation is essentially copied from peer.py. The code
in peer.py will eventually be deleted.
Differential Revision: https://phab.mercurial-scm.org/D334
There are a well-defined set of commands constituting the wire
protocol. Interaction with these and methods for calling them in
batches are exposed via methods on peer instances.
Let's formalize support for these features in abstract classes.
The command parts come from the existing wireproto.wirepeer class.
The batch methods come from peer.peerrepository.
Ample documentation has been added as part of defining the interfaces.
# no-check-commit
Differential Revision: https://phab.mercurial-scm.org/D333
There are various interfaces for interacting with repositories
and peers. They form a contract for how one should interact with
a repo or peer object.
The contracts today aren't very well-defined or enforced. There
have been several bugs over the years where peers or repo types
have forgotten to implement certain methods. In addition, the
inheritance of some classes is wonky. For example, localrepository
doesn't inherit from an interface and the god-object nature of
that class means the repository interface isn't well-defined. Other
repository types inherit from localrepository then stub out
methods that don't make sense (e.g. statichttprepository
re-defining locking methods to fail fast).
Not having well-defined interfaces makes implementing alternate
storage backends, wire protocol transports, and repository types
difficult because it isn't clear what exactly needs to be
implemented.
This patch starts the process of attempting to establish more
order to the type system around repositories and peers.
Our first patch starts with a problem space that already has a
partial solution: peers. The peer.peerrepository class already
somewhat defines a peer interface. But it is missing a few things
and the total interface isn't well-defined because it is combined
with wireproto.wirepeer.
Our newly-established basepeer class uses the abc module to
declare an abstract base class with the properties and methods that
a generic peer must implement.
We create a new class that inherits from it. This class will hold
our other future abstract base classes / interfaces so we can expose
a unified base class/interface.
We don't yet use the new interface because subsequent additions
will break existing code without some refactoring first.
A new module (repository.py) was created to hold the interfaces.
I could have put things in peer.py. However, I have plans to
eventually add interfaces to define repository and storage types.
These almost certainly require a new module. And I figured having
all the interfaces live in one module makes sense. So I created
repository.py to be that future home.
Differential Revision: https://phab.mercurial-scm.org/D332
a3022f57803b made `util.nogc` a no-op. It was to optimize PyPy. But it slows
down CPython if many objects (like 300k+) are created.
For example, running `hg log -r .` without extensions in `hg-committed` with
14k+ obsmarkers have the following times:
before | after
hg | chg | hg | chg
-----------------------------
1.262 | 0.860 | 1.077 | 0.619 (seconds, best of 20 runs)
Therefore let's re-enable nogc for CPython.
Differential Revision: https://phab.mercurial-scm.org/D402
troubled has been renamed into isunstable but troubled was calling unstable
instead. Fix the mistake.
Differential Revision: https://phab.mercurial-scm.org/D384
We have at least three types with a close() and a release() method
where the close() method is supposed to be called on success and the
release() method is supposed to be called last, whether successful or
not. Two of them (transaction and dirstateguard) already have
identical implementations of __enter__ and __exit__. Let's extract a
base class for this, so we reuse the code and so the third type
(transactionmanager) can also be used as a context manager.
Differential Revision: https://phab.mercurial-scm.org/D392
The transactionmanager() constructor just assigned a few variables and
cannot fail, so it's safe to move it inside the earlier try/except.
Differential Revision: https://phab.mercurial-scm.org/D391
Since 6f17bd68a306 (exchange: drop support for lock-based unbundling
(BC), 2017-08-06), there is no more remote locking.
Differential Revision: https://phab.mercurial-scm.org/D390
The variable has been used only within a single function since
5d683cc9670f (push: elevate phase transaction to cover entire
operation, 2014-11-21), so there's no need to keep it on the pushop
object.
Differential Revision: https://phab.mercurial-scm.org/D389
morestatus extension in fbext use to show more context about the state of the
repo like the repository is in a unfinished merge state, or a rebase is going
on, or histedit is going on, listing the files which need to be resolved and
also suggesting ways to handle the situation.
This patch moves the extension directly to core by plugging it into the
--verbose flag of the status command. So now if you are in any unfinished state
and you do hg status -v, it will show you details and help related to the state.
The extension in fbext also shows context about unfinished update state
which is not ported to core as that plug in hooks to update command which need
to be tackled somewhat differently.
The following configuration will turn the behaviour on by default
[commands]
status.verbose = 1
You can also skip considering some states like bisect as follows:
[commands]
status.skipstates=bisect
This patch also adds test for the feature.
.. feature::
``hg status -v`` can now show unfinished state. For example, when in
an unfinished rebase state, ``hg status -v`` might show::
# The repository is in an unfinished *rebase* state.
# No unresolved merge conflicts.
# To continue: hg rebase --continue
# To abort: hg rebase --abort
Differential Revision: https://phab.mercurial-scm.org/D219
Changeset aa97e972460f introduce more complex logic around
'bundleoperation.gettransaction'. In that process it turns the old "attribute"
into a proper method which breaks the code that detects the "transaction
availability".
The change was visible in 'test-acl.t', fixing this reverts the test changes.
Differential Revision: https://phab.mercurial-scm.org/D303
Peer types are supposed to conform to a formal interface defined by
peer.peerrepository and wireproto.wirepeer. Every "public" attribute on
*peer types makes it harder to understand what attributes are part
of the interface and what are instance specific.
This commit converts a number of "public" instance attributes and
methods on sshpeer to internal so they can't be confused to be part of
the peer API.
The URL-related instance attributes were introduced in 904c418bea16
in 2005. AFAICT most of them aren't used and could potentially be
removed. But I kept them around anyway.
I also reorded some code to make things slightly easier to read.
.. api::
Rename attributes on sshpeer to reflect peer API
Differential Revision: https://phab.mercurial-scm.org/D331
The last use of this API was removed in 3bcb9f9a4a63 in 2016. While
not formally deprecated, as of the last commit the code is no longer
explicitly tested. I think the new API has existed long enough for
people to transition to it.
I also have plans to more formalize the peer API and removing batch()
makes that work easier.
I'm not convinced the current client-side API around batching is
great. But it's the best we have at the moment.
.. api:: remove peer.batch()
Replace with peer.iterbatch().
Differential Revision: https://phab.mercurial-scm.org/D320
The remote batching code is difficult to read. Let's improve it.
As part of the refactor, the future returned by method calls on
batchiter() instances is now populated. However, you still need to
consume the results() generator for the future to be set. But at
least now we can stuff the future somewhere and not have to worry
about aligning method call order with result order since you can
use a future to hold the result.
Also as part of the change, we now verify that @batchable generators
yield exactly 2 values. In other words, we enforce their API.
The non-iter batcher has been unused since 3bcb9f9a4a63. And to my
surprise we had no explicit unit test coverage of it! test-batching.py
has been overhauled to use the iterating batcher.
Since the iterating batcher doesn't allow non-batchable method
calls nor local calls, tests have been updated to reflect reality.
The iterating batcher has been used for multiple releases apparently
without major issue. So this shouldn't cause alarm.
.. api::
@peer.batchable functions must now yield exactly 2 values
Differential Revision: https://phab.mercurial-scm.org/D319
@peer.batchable decorated generator functions have two forms:
yield value, None
and
yield args, future
yield value
These forms have been present since the decorator was introduced.
There are currently no in-repo consumers of the first form. So this
commit removes support for it.
Note that remoteiterbatcher.submit() asserts the 2nd form. And
3bcb9f9a4a63 removed the last user of remotebatcher, forcing everyone
to remoteiterbatcher. So anything relying on this in the wild would
have been broken since 3bcb9f9a4a63.
.. api::
@peer.batchable can no longer emit local values
Differential Revision: https://phab.mercurial-scm.org/D318
remoteiterbatcher (unlike remotebatcher) only supports batchable
commands. This claim can be validated by comparing their
implementations of submit() and noting how remoteiterbatcher assumes
the invoked method has a "batchable" attribute, which is set by
@peer.batchable.
remoteiterbatcher has a custom __getitem__ that was trying to
validate that only batchable methods are called. However, it was only
validating that the called method exists, not that it is batchable.
This wasn't a big deal since remoteiterbatcher.submit() would raise
an AttributeError attempting to `mtd.batchable(...)`.
Let's fix the check and convert it to ProgrammingError, which may
not have been around when this was originally implemented.
Differential Revision: https://phab.mercurial-scm.org/D317
This extracts charencode.c from parsers.c, which seems big enough for me
to hesitate to add new JSON functions. Still charencode.o is linked to
parsers.so to avoid duplication of binary codes.
As we changed the meaning of unstable between the old vocabulary and the new
one, we can't reuse the unstable method name at the risk of breaking
extensions calling unstable and getting a wrong result.
Instead rename troubled into isunstable so extensions will continue to work.
The renaming is done according to
https://www.mercurial-scm.org/wiki/CEDVocabulary.
Differential Revision: https://phab.mercurial-scm.org/D242
Rename troubles context method into instabilities.
Copy the old troubles method and add a deprecation warning. This way
extensions calling troubles will see the deprecation warning but will not
break due to new return values.
The renaming is done according to
https://www.mercurial-scm.org/wiki/CEDVocabulary.
Differential Revision: https://phab.mercurial-scm.org/D238
This moves manifest stripping to a separate function so implementations of the
manifest that don't support stripping can replace this function with a no-op.
I considered adding a strip api to the manifestlog, so other implementations
could make it a no-op there, but it seems like strip might be unique to the
revlog implementation, and therefore shouldn't be present on the generic api.
Differential Revision: https://phab.mercurial-scm.org/D292
This refactors broken linkrev collection such that manifest collection is in a
separate function. This allows extensions to replace the manifest collection
with a non-revlog oriented version.
I considered moving the collect changes function onto the manifestlog itself, so
it would be behind the abstraction, but since the store we're building doesn't
even have the concept of strip, embeding that concept in the manifestlog api
seemed odd.
Differential Revision: https://phab.mercurial-scm.org/D291
I'm not sure this is right, since this should either be bytes or str
to match what's going on in the revlog layer.
Differential Revision: https://phab.mercurial-scm.org/D271
I'm not sure if this is a realistic problem, but doing this avoids
some pretty awful test failures on Python 3, and it looks like it
should be harmless.
Differential Revision: https://phab.mercurial-scm.org/D274
Fixes some Python 3 regressions.
We don't use %d here because the part id is actually an
Optional[int]. It should always be initialized to a non-None value by
the time this code executes, but we shouldn't blindly depend on that
being the case.
Differential Revision: https://phab.mercurial-scm.org/D272
The changelog object can store recently added revisions in memory
until the transaction is committed. We don't want to lose those
changes even if repo.invalidate(clearfilecache=True), so let's skip
the changelog when it has such 'delayed' changes.
Differential Revision: https://phab.mercurial-scm.org/D152
Peer instances are supposed to conform to a well-defined API so
consumers can be agnostic about the underlying peer type.
To reinforce this, this commit renames a handful of instance
attributes on httpeer so they no longer have a "public" name.
Differential Revision: https://phab.mercurial-scm.org/D268
The previous changeset removed the last consumer of requirements. I'm
not sure when supportedformats became unused. But I couldn't find
any obvious instances where it is being used. It likely stems from
peers being derived from repository instances several years ago and
is a holdover from that day.
Differential Revision: https://phab.mercurial-scm.org/D266
As part of formalizing the peer interface, I audited for attribute
accesses for non-internal names to find API violations. This
uncovered the code changed in this commit.
localpeer.requirements is just an alias to the repo's requirements
attribute. So, change the code to get the data from the source
instead of relying on a one-off attribute in the localpeer type.
Differential Revision: https://phab.mercurial-scm.org/D265
Locking over the wire protocol and the "addchangegroup" wire
protocol command has been deprecated since f8e443eb02c9, which was
first part of Mercurial 0.9.1.
Support for handling these commands from sshserver was dropped in
93297d5f4df2 in 2015, effectively locking out pre 0.9.1 clients
from new servers.
However, client-side code for calling lock and addchangegroup is
still present in exchange.py and the various peer classes to
facilitate pushing to pre 0.9.1 servers.
The lock-based pushing mechanism is extremely brittle. 0.9.1 was
released in July 2006 and I highly doubt anyone is still running
such an ancient version of Mercurial on a server. I'm about to
refactor the peer API and I don't think it is worth keeping
support for this ancient protocol feature. So, this commit removes
client support for the lock-based pushing mechanism. This means
modern clients will no longer be able to push to pre 0.9.1 servers.
Differential Revision: https://phab.mercurial-scm.org/D264
This patch also adds extra loading entry for internal merge tools to
extensions.py, for similarity to other decorators defined in
registrar.py.
This patch uses "internalmerge" for decorator class name, instead of
original "internaltool", because the latter is too generic.
BTW, after this patch, 4-spaces indentation is added to the 1st line
of internal merge tool description docstring, and this may make
already translated entries in *.po fuzzy.
Even though this indentation is required for "definition list" in reST
syntax, absence of it has been overlooked, because help.makeitemsdoc()
forcibly inserts it at generation of online help.
But this forcible insertion causes formatting issue (I'll send another
patch series for this). Therefore, this additional indentation should
be reasonable.
Let "isignoreddir" function first check that supplied directory is itself
ignored before walking recursively into its content. Otherwise, the command is
awfully slow when one has an ignored directory with a lot of content.
Update and rephrase function docstring accordingly.
pushvars extension in fbext adds a --pushvars flag to push command using which
one send strings to server which becomes environment variables there prepended
with HG_USERVAR_. These variables can then be used to run hooks on the server.
The extension is moved directly to core and unbundling of the strings and
converting them to environment variables at server is disabled by default for
security reasons. One can turn that on by following config:
[push]
pushvars.server = true
This patch also adds the test for the extension.
Differential Revision: https://phab.mercurial-scm.org/D210
Rename bumped to phase-divergent in all external user-facing output. Only
update user-facing output for the moment, variables names, templates keyword
and potentially configuration would be done in later series.
The renaming is done according to
https://www.mercurial-scm.org/wiki/CEDVocabulary.
Differential Revision: https://phab.mercurial-scm.org/D216
Rename divergent to content-divergent in all external user-facing output. Only
update user-facing output for the moment, variables names, templates keyword
and potentially configuration would be done in later series.
The renaming is done according to
https://www.mercurial-scm.org/wiki/CEDVocabulary.
Differential Revision: https://phab.mercurial-scm.org/D215
This commit makes it so sparse treats passed paths as CWD-relative,
not repo-root-realive. This is a more intuitive behavior in my (and some
other FB people's) opinion.
This is breaking change however. My hope here is that since sparse is
experimental, it's ok to introduce BCs.
The reason (glob)s are needed in the test is this: in these two cases we
do not supply path together with slashes, but `os.path.join` adds them, which
means that under Windows they can be backslashes. To demonstrate this behavior,
one could remove the (glob)s and run `./run-tests.py test-sparse.t` from
MinGW's terminal on Windows.
Current logic is misleading (it says it drops only absolute paths, but
it actually drops all of them), not cross-platform (does not support Windows)
and IMO just wrong (as it should just error out if absolute paths are given).
This commit fixes it.
If the user sets color.mode=terminfo, and then runs in the shell inside of emacs
(so TERM=dumb), the previous behavior was that it would warn about no terminfo
entry for setab/setaf, and then warn about 'failed to set color mode to
terminfo'. The first warning is silenced by carrying 'formatted' through to
_terminfosetup, the second is silenced by using 'formatted' instead of
ui.formatted().
If --color=on (or ui.color=always) is specified, this will still warn, since the
formatted boolean is set to true in these cases.
Differential Revision: https://phab.mercurial-scm.org/D223
Rename unstable to orphan in all external user-facing output. Only update
user-facing output for the moment, variables names, templates keyword and
potentially configuration would be done in later series.
The renaming is done according to
https://www.mercurial-scm.org/wiki/CEDVocabulary.
Differential Revision: https://phab.mercurial-scm.org/D214
Rename trouble(s) to instability in all external user-facing output. Only
update user-facing output for the moment, variables names, templates keyword
and potentially configuration would be done in later series.
The renaming is done according to
https://www.mercurial-scm.org/wiki/CEDVocabulary.
Differential Revision: https://phab.mercurial-scm.org/D213
When a transaction is started, we must load the hookargs from the
bundleoperation object to the transaction so that they can be used in the
transaction. Also this patch makes sure no more hookargs are added to the
bundleoperation object once the transaction starts.
This is a part of porting fb extension bundle2hooks to core.
Differential Revision: https://phab.mercurial-scm.org/D209
There are extensions like pushrebase, pushvars which run hooks on a server
before taking the lock. Since the lock is not taken, transaction is not there,
so the hookargs can't be stored on the transaction. Adding hooksargs to bundle
operation object will help in running hooks before taking the lock.
This is a part of moving fb's extension bundle2hooks to core.
Differential Revision: https://phab.mercurial-scm.org/D208
This was previously landed as 4bc0c14fb501 but backed out in b63351f6a2 because
it broke hooks mid-rebase and caused conflict resolution data loss in the event
of unexpected exceptions. This new version adds the behavior back but behind a
config flag, since the performance improvement is notable in large repositories.
The old commit message was:
Recently we switched rebases to run the entire rebase inside a single
transaction, which dramatically improved the speed of rebases in repos with
large working copies. Let's also move the dirstate into a single dirstateguard
to get the same benefits. This let's us avoid serializing the dirstate after
each commit.
In a large repo, rebasing 27 commits is sped up by about 20%.
I believe the test changes are because us touching the dirstate gave the
transaction something to actually rollback.
(grafted from 9e3dc3a1638b9754b58a0cb26aaa75d868058109)
(grafted from 7d38b41d2266d9a02a15c64229fae0da5738dcec)
Differential Revision: https://phab.mercurial-scm.org/D135
This simplifies the method slightly. It does create a full list of
paths while doing so, but it's not a lot of data anyway (besides, I
would think references to strings are no larger than (references to?)
True).
mergestate.unresolved() is a generator, so it seems better for it to
rely on iteritems() than items(), although it also seems unlikely for
it to make a noticeable difference.
I don't know why applying an empty changegroup should be an error. It
seems harmless. I suspect the check was there to find code that
creates empty changegroups just because that would be wasteful. Let's
use develwarn() for that instead, so we catch any such cases that run
with our test runner, but we still allow others to generate empty
changegroups if they want to.
We have run into this check at Google once or twice and had to work
around it, but I'm changing this not so much because of that, but
because it seems like it shouldn't be an error.
I also changed the message slightly to be more modern ("changelog
group" -> "changegroup") and more generic ("received" -> "applied").
Applying an empty changegroup has been an error since the
beginning. The only exception was strip, which would allow to apply an
empty changegroup from the temporary bundle. However, the emptyok=True
option was only set for bundle1 bundles. In other words, temporary
bundle2 bundles would fail if they were empty.
Bundle2 has now been used enough that it seems safe to say that we
simply don't create bundle2 bundles with empty changegroups. That also
suggests that we never create bundle1 bundles with empty changegroups
(i.e. empty bundle1 bundles, since bundle1 is just a changegroup),
because, AFAICT, the code leading up to the application of the bundle
is the same for bundle1 and bundle2.
Therefore, let's stop passing emptyok=True, so we more clearly get the
same behavior for bundle1 and bundle2.
The "patterns"/"include" in "patternspat"/"includepat" is redundant,
so drop it. Also a "_" prefix since it's "private".
Inline the "pm"/"im" variables.
The code was refactored during the move to be more procedural
instead of using string formatting. This has the benefit of not
writing empty sections, which changed tests.
The sparse extension maintains caches for the sparse files
to a signature and a signature to a matcher. This allows the
sparse matchers to be resolved quickly, which is apparently
something that can occur in loops.
This patch ports the sparse caches to the localrepo class
pretty much as-is. There is potentially room to improve the
caching mechanism. But that can be done as a follow-up.
The default invalidatecaches() now clears the relevant sparse
cache. invalidatesignaturecache() has been moved to sparse.py.
This method is reasonably well-contained and simple to move.
As part of the move, some light formatting was performed.
A "working copy" reference in an error message was changed to
"working directory."
The biggest change was to _refreshoncommit() in sparse.py. It
was previously checking for the existence of an attribute on
the repo instance. Since the moved function now returns empty
data if sparse isn't enabled, we unconditionally call the
new function. However, we do have to protect another method
call in that function. This will all be unhacked eventually.
Currently, the sparse extension sniffs repo instances for
attributes defined by the sparse extension to determine if
sparse is enabled. As we move code away from repo instances,
these checks will be a bit more brittle.
We introduce a module-level variable to track whether sparse is
enabled as a temporary workaround.
One more step towards weaning off methods on repo instances and
moving code to core. While this function is only used once and
is simple, it needs to exist on its own so Facebook can monkeypatch
it to enable simplecache integration.
This patch marks the beginning of moving code from the sparse
extension into core. The goal is to move as much of the
functionality as possible into core, where it will be an
experimental feature. The extension will likely continue to
exist to enable the feature and provide UI elements.
As part of the move, the repo method was converted to a module
function. It doesn't need to exist on repos.
An error message was also updated to reflect that an error isn't
necessarily from the .hg/sparse file. The API should be updated
later to pass in a filename so the error can be more descriptive.
Copyright of the added file was copied from the sparse extension.
In blockdescendants(), we had an assertion when line range of a merge
changeset was not consistent depending on which parent was considered for
computation. For instance, this might occur when file content (in lookup
range) is significantly different between parent branches of the merge as
demonstrated in added tests (where we almost completely rewrite the "baz" file
while also introducing similarities with its content in the other branch we
later merge to).
Now, in such case, we combine line ranges from all parents by storing the
envelope of both line ranges. This is conservative (the line range is
extended, possibly unnecessarily) but at least this should avoid missing
descendants with changes in a range that would fall in that of one parent but
not in another one (the case of "baz: narrow change (2->2+)" changeset in
tests).
Switch the lone call in merge.py to use it.
As with past refactors, the goal is to make wctx hot-swappable with an
in-memory context in the future. This change should be a no-op today.
Now, files (to be truncated) are opened with checkambig=True, only if
localrepository caches it.
Therefore, straightforward "copy if EPERM" is always reasonable to
avoid file stat ambiguity at closing.
This patch makes checkambigatclosing close wrapper copy the target
file, and advance mtime on it after renaming, if EPERM. This can avoid
file stat ambiguity, even if the target file is owned by another (see
issue5418 and issue5584 for detail).
This patch factors main logic out instead of changing
checkambigatclosing._checkambig() directly, in order to reuse it.
Now, transaction can determine whether avoidance of file stat
ambiguity is needed for each files, by blacklist "checkambigfiles".
For similarity to truncation in _playback(), this patch apply
checkambig=True only on limited files in code paths below.
- restoring files by util.copyfile(), in _playback()
(checkambigfiles itself is examined at first, because it as a
keyword argument might be None)
- writing files at finalization of transaction, in _generatefiles()
This patch reduces cost of checking stat at writing out and restoring
files, which aren't filecache-ed.
Advancing mtime by os.utime() fails for EPERM, if the target file is
owned by another. 0d920bcb0fd1 and related changes made some code
paths give advancing mtime up in such case, to fix issue5418.
This causes file stat ambiguity (again), if it is owned by another.
https://www.mercurial-scm.org/wiki/ExactCacheValidationPlan
To avoid file stat ambiguity in such case, especially for
.hg/dirstate, c75c7b3e3284 made vfs.rename() copy the target file, and
advance mtime of renamed one again, if EPERM (see issue5584 for detail).
But straightforward "copy if EPERM" isn't reasonable for truncation of
append-only files at rollbacking, because rollbacking might cost much
for truncation of many filelogs, even though filelogs aren't
filecache-ed.
Therefore, this patch introduces blacklist "checkambigfiles", and
avoids file stat ambiguity only for files specified in this blacklist.
This patch consists of two parts below, which should be applied at
once in order to avoid regression.
- specify 'checkambig=True' at vfs.open(mode='a') in _playback()
according to checkambigfiles
- invoke _playback() with checkambigfiles
- add transaction.__init__() checkambigfiles argument, for _abort()
- make localrepo instantiate transaction with _cachedfiles
- add rollback() checkambigfiles argument, for "hg rollback/recover"
- make localrepo invoke rollback() with _cachedfiles
After this patch, straightforward "copy if EPERM" will be reasonable
at closing the file opened with checkambig=True, because this policy
is applied only on files, which are listed in blacklist
"checkambigfiles".
This information is used to make transaction handle these files
specially, in order to avoid file stat ambiguity of them.
Gathering information about cached files via annotation classes can
avoid overlooking properties newly introduced in the future.
Add a 'successorssets' template that returns the list of all closest known
sucessorssets for a changectx. The elements of the list are changesets.
The "closest successors" are the first locally known revisions encountered
while, walking successors markers. It uses successorsets previously modified
to support the closest argument.
This logic respect repository filtering. So hidden revision will be skipped by
this logic unless --hidden is specified. Since we only display the visible
predecessors, this template will not display anything in most case. It makes a
good candidate for inclusion in the default log output.
I updated the test-obsmarker-template.t test file introduced with the
predecessors template to test successorssets template.
Add a closest argument to successorssets changing the definition of latest
successors.
With "closest=false" (current behavior), latest successors are "leafs" on the
obsmarker graph. They don't have any successor and are known locally.
With "closest=true", latest successors are the closest locally-known
changesets that are visible in the repository or repoview. Closest successors
can be then obsolete, orphan.
This will be used in a later patch to show the closest successor of
changesets with the successorssets template.
We plan to add a new argument to successorsets. But first we need to update
all callers to pass cache argument explicitly to avoid arguments confusion.
Clarify successorssets documentation before we start updating the main function.
This patch serie will introduce the successorssets template, the opposite of
predecessor templates.
Successors will use successorssets function and requires some improvement so
before doing that, we clean up successorssets a bit.
Previously there is a suspicious `if False and delta > 0` which dates back
to the beginning of hgext/record.py (f995f03023c7).
The "trimming context lines" feature could be useful (and is used by the
next patch). So let's enable it. This patch adds a new `maxcontext`
parameter to `recordhunk` and `parsepatch`, changing the `if False`
condition to respect it.
The old `trimcontext` implementation is also wrong - it does not update
`toline` correctly and it does not do the right thing for `before` context.
A doctest was added to guard us from making a similar mistake again.
Since `maxcontext` is set to `None` (unlimited), there is no behavior
change.
There are no remaining users of 'mustaudit' so we can safely drop the API.
External user are unlikely from a quick research so no deprecation is added.
The whole 'mustaudit' API is quite complex compared to its actual usage by its
unique user in stream clone.
Instead we add a "auditpath" parameter to 'vfs.__call_'. The stream clone code
then explicitly open files with path auditing disabled.
The 'mustaudit' API will be cleaned up in the next changeset.
This is a first basic visible usage of the changes tracking in the transaction.
We adds a new function computing the pre-existing changesets obsoleted by a
transaction and a transaction call back displaying this information.
Example output:
added 1 changesets with 1 changes to 1 files (+1 heads)
3 new obsolescence markers
obsoleted 1 changesets
The goal is to evolve the transaction summary into something bigger, gathering
existing output there and adding new useful one. This patch is a good first step
on this road. The new output is basic but give a user to the content of
tr.changes['obsmarkers'] and give an idea of the new options we haves. I expect
to revisit the message soon.
The caller recording the transaction summary should also be moved into a more
generic location but further refactoring is needed before it can happen.
The obsstore collaborate with transaction to make sure we track all the
obsmarkers added during a transaction. This will be useful for various usages:
hooks, caches, better output, etc.
This is the seconds kind of data added to tr.changes (first one was added revisions)
Before this patch, immediate value is assigned to dirstate._cwd, if
ui.forcecwd is specified at instantiation of dirstate.
But this doesn't work as expected in some cases.
For example, hgweb set ui.forcecwd after instantiation of repo object.
If an extension touches repo.dirstate in its reposetup(), dirstate is
instantiated without setting ui.forcecwd, and dirstate.getcwd()
returns incorrect result.
In addition to it, hgweb.__init__() can take already instantiated repo
object, too. In this case, repo.dirstate might be already
instantiated, even if all enabled extensions don't so in their own
reposetup().
To avoid such issue, this patch centralizes _cwd handling into _cwd
method.
This issue can be reproduced by running test-hgweb-commands.t with
fsmonitor-run-tests.py.
The general delta heuristic to select a delta do not scale with the number of
branch. The delta base is frequently too far away to be able to reuse a chain
according to the "distance" criteria. This leads to insertion of larger delta (or
even full text) that themselves push the bases for the next delta further away
leading to more large deltas and full texts. This full text and frequent
recomputation throw Mercurial performance in disarray.
For example of a slightly large repository
280 000 files (2 150 000 versions)
430 000 changesets (10 000 topological heads)
Number below compares repository with and without the distance criteria:
manifest size:
with: 21.4 GB
without: 0.3 GB
store size:
with: 28.7 GB
without 7.4 GB
bundle last 15 00 revisions:
with: 800 seconds
971 MB
without: 50 seconds
73 MB
unbundle time (of the last 15K revisions):
with: 1150 seconds (~19 minutes)
without: 35 seconds
Similar issues has been observed in other repositories.
Adding a new option or "feature" on stable is uncommon. However, given that this
issues is making Mercurial practically unusable, I'm exceptionally targeting
this patch for stable.
What is actually needed is a full rework of the delta building and reading
logic. However, that will be a longer process and churn not suitable for stable.
In the meantime, we introduces a quick and dirty mitigation of this in the
'experimental' config space. The new option introduces a way to set the maximum
amount of memory usable to store a diff in memory. This extend the ability for
Mercurial to create chains without removing all safe guard regarding memory
access. The option should be phased out when core has a more proper solution
available.
Setting the limit to '0' remove all limits, setting it to '-1' use the default
limit (textsize x 4).
The general delta heuristic to select a delta do not scale with the number of
branch. The delta base is frequently too far away to be able to reuse a chain
according to the "distance" criteria. This leads to insertion of larger delta (or
even full text) that themselves push the bases for the next delta further away
leading to more large deltas and full texts. This full text and frequent
recomputation throw Mercurial performance in disarray.
For example of a slightly large repository
280 000 files (2 150 000 versions)
430 000 changesets (10 000 topological heads)
Number below compares repository with and without the distance criteria:
manifest size:
with: 21.4 GB
without: 0.3 GB
store size:
with: 28.7 GB
without 7.4 GB
bundle last 15 00 revisions:
with: 800 seconds
971 MB
without: 50 seconds
73 MB
unbundle time (of the last 15K revisions):
with: 1150 seconds (~19 minutes)
without: 35 seconds
Similar issues has been observed in other repositories.
Adding a new option or "feature" on stable is uncommon. However, given that this
issues is making Mercurial practically unusable, I'm exceptionally targeting
this patch for stable.
What is actually needed is a full rework of the delta building and reading
logic. However, that will be a longer process and churn not suitable for stable.
In the meantime, we introduces a quick and dirty mitigation of this in the
'experimental' config space. The new option introduces a way to set the maximum
amount of memory usable to store a diff in memory. This extend the ability for
Mercurial to create chains without removing all safe guard regarding memory
access. The option should be phased out when core has a more proper solution
available.
Setting the limit to '0' remove all limits, setting it to '-1' use the default
limit (textsize x 4).
People often want to know what they are working on *now*. As part of
this, they also commonly want to know how that work is related to other
changesets in the repo so they can perform common actions like rebase,
histedit, and merge.
`hg show work` made headway into this space. However, it is geared
towards a complete repo view as opposed to just the current line of
work. If you have a lot of in-flight work or the repo has many heads,
the output can be overwhelming. The closest thing Mercurial has to
"show me the current thing I'm working on" that doesn't require custom
revsets is `hg qseries`. And this requires MQ, which completely changes
workflows and repository behavior and has horrible performance on large
repos. But as sub-optimal as MQ is, it does some things right, such as
expose a model of the repo that is easy for people to reason about.
This simplicity is why I think a lot of people prefer to use MQ, despite
its shortcomings.
One common development workflow is to author a series of linear
changesets, using bookmarks, branches, anonymous heads, or even topics
(3rd party extension). I'll call this a "stack." You periodically
rewrite history in place (using `hg histedit`) and reparent the stack
against newer changesets (using `hg rebase`). This workflow can be
difficult because there is no obvious way to quickly see the current
"stack" nor its relation to other changesets. Figuring out arguments to
`hg rebase` can be difficult and may require highlighting and pasting
multiple changeset nodes to construct a command.
The goal of this commit is to make stack based workflows simpler
by exposing a view of the current stack and its relationship to
other releant changesets, notably the parent of the base changeset
in the stack and newer heads that the stack could be rebased or merged
into.
Introduced is the `hg show stack` view. Essentially, it finds all
mutable changesets from the working directory revision in both
directions, stopping at a merge or branch point. This limits the
revisions to a DAG linear range.
The stack is rendered as a concise list of changesets. Alongside the
stack is a visualization of the DAG, similar to `hg log -G`.
Newer public heads from the branch point of the stack are rendered
above the stack. The presence of these heads helps people understand
the DAG model and the relationship between the stack and changes made
since the branch point of that stack. If the "rebase" command is
available, a `hg rebase` command is printed for each head so a user
can perform a simple copy and paste to perform a rebase.
This view is alpha quality. There are tons of TODOs documented
inline. But I think it is good enough for a first iteration.
The ignore regular expression has been updated to detect
"inconsistent config." If present, we track which configs have
that set and we suppress the conflicting defaults error for those
options.
I also added named groups to the regexp to aid readability.
A comment was added to profiling.py to make a desired inconsistent
value error go away.
It isn't needed that storecache is derived from repofilecache.
Changes in this patch allow repofilecache and storecache to do in own
__init__() differently from each other.
This is a fix for my 1c4d8237ba73, which used 'bool(dir)' as
checkambig value for revlog.__init__().
I can't remember why I did so in 1c4d8237ba73, but this is obviously
wrong, because only root indexfile is cached via filecache-ed property
of localrepository.
* Scope of "value" is reduced
* index_baserev() is documented
* Error is no longer redundantly set for -2 return values
* Error values are compared <= -2 instead of == -2 to protect
against odd failure scenarios
I've seen revlog._deltachain() appear in a number of performance
profiles. I suspect there are 2 reasons for this:
1. Delta chain resolution performs many index lookups, thus triggering
population of index tuples. Creating possibly tens of thousands of
PyObject will have overhead.
2. Delta chain resolution is a tight loop.
By moving delta chain resolution to C, we can defer instantiation
of full index entry tuples and make the loop faster courtesy of
not running in Python.
We can measure the impact to delta chain resolution via
`hg perflogrevision` using the mozilla-central repo with a recent
manifest having delta chain length of 33726:
$ hg perfrevlogrevision -m 364895
! full
! wall 0.367585 comb 0.370000 user 0.340000 sys 0.030000 (best of 27)
! wall 0.357581 comb 0.360000 user 0.350000 sys 0.010000 (best of 28)
! deltachain
! wall 0.010644 comb 0.010000 user 0.010000 sys 0.000000 (best of 270)
! wall 0.000292 comb 0.000000 user 0.000000 sys 0.000000 (best of 8729)
$ hg perfrevlogrevision --cache -m 364895
! deltachain
! wall 0.003904 comb 0.000000 user 0.000000 sys 0.000000 (best of 712)
! wall 0.000284 comb 0.000000 user 0.000000 sys 0.000000 (best of 9926)
The first test measures savings from both not instantiating index
entries and moving to C. The second test (which doesn't clear the
index caches) essentially isolates the benefits of moving from Python
to C. It still shows a 13.7x speedup (versus 36.4x). And there are
multiple milliseconds of savings within the critical path for resolving
revision data. I think that justifies the existence of C code.
A more striking example of the benefits of this change can be
demonstrated by timing `hg debugdeltachain -m` for the mozilla-central
repo:
$ time hg debugdeltachain -m > /dev/null
before: 1057.4s
after: 503.3s
PyPy2.7 5.8.0: 220.0s
It's worth noting that the C code isn't as optimal as it could be.
We're still instantiating a new PyObject for every revision. A future
optimization would be to reuse the PyObject on the cached index tuple.
We could potentially also get wins by using a memory array of raw
integers. There is also room for a delta chain cache on revlog
instances. Of course, the best optimization is to implement revlog
reading outside of Python so Python doesn't need to be concerned
about the relatively expensive index entries and operations on them.
Let us start registering the existing option. I'm starting with the 'devel'
section because it is full of useful things that are poorly documented. So
registering these will more than other section.
The function lost its last caller in ae209b610844 (patch: replace
functions in fsbackend to use vfs, 2014-06-05) when the callers
started relying on the opener to do the join.
patchbackend() seems to call it on an arbitrary backend, so it seems
to be part of the API. Since all subclasses do something in their
close() methods, I decided to let this one raise an exception rather
than just pass.
Extensions sometimes wants to add other information in the default log output
format (when no templating is used).
Add an empty function named '_exthook' for easing the extension life.
Extensions will be able to wrap this function and collaborate to display
additional information.
Exthook is called after displaying troubles and just before displaying the
files, extra and description.
Add a new test file to test it and not pollute other test files.
One hitch is that sometimes fcd is actually an absentfilectx which does not
expose any mutator functions. In order to still use the context functions,
we look up the underlying workingfilectx to perform the write there.
One alternate way would be to put the write functions on the absentfilectx and
have them pass-through. While this makes the callsites cleaner, we would need
to decide what its getter functions would return after this point, since
returning None for `data` (and True for `isabsent()`) might no longer be
correct after a write. I discussed with Sidd about just having the getters
raise RuntimeErrors after a mutator has been called, but we actually call
isabsent() in merge.py after running the internal merge tools.
Yuya pointed out that using mutable value as the default could be problematic.
To work around this we now support callable object as default value. This
allows for creating new mutable objects on demand when needed.
None of this function has been used in the past 5 years, so I think it is safe
to just kill them. All code accessing rich markers is using 'getmarkers(...)'
instead (or raw markers).
Extensions can have a 'configtable' mapping and use
'registrar.configitem(table)' to retrieve the registration function.
This behave in the same way as the other way for extensions to register new
items (commands, colors, etc).
We simplify the unstable computation code, skipping the expensive creation of
changectx object. We focus on efficient set operation and revnumber centric
functions.
In my mercurial development repository, this provides a 3x speedup to the
function:
before: 5.319 ms
after: 1.844 ms
repo details:
total changesets: 40886
obsolete changesets: 7756
mutable (not obsolete): 293
unstable: 30
There are cases in opts handling in debugcommands.py where we don't need to
convert opts keys back to bytes as there are some handful cases and no other
function using opts value. Using r'', we prevent the transformer to add
a b'' which will keep the value str.
Otherwise nested template formatter would not see the context objects.
It's just a boolean flag now. We might want to change it to 'ctxs -> items'
function so changectx attributes are populated automatically in JSON, but
I'm not sure.
It's now common that an old node gets replaced by zero or more new nodes,
that could happen with amend, rebase, histedit, etc. And it's a common
requirement to do bookmark movements, strip or obsolete nodes and even
moving working copy parent.
Previously, amend, rebase, history have their own logic doing the above.
This patch is an attempt to unify them and future code.
This enables new developers to be able to do "replace X with Y" thing
correctly, without any knowledge about bookmarks, strip or obsstore.
The next step will be migrating rebase to the new API, so it works inside a
transaction, and its code could be simplified.
For long, the fact that strip does not work inside a transaction and some
code has to work with both obsstore and fallback to strip lead to duplicated
code like:
with repo.transaction():
....
if obsstore:
obsstore.createmarkers(...)
if not obsstore:
repair.strip(...)
Things get more complex when you want to call something which may call strip
under the hood. Like you cannot simply write:
with repo.transaction():
....
rebasemod.rebase(...) # may call "strip", so this doesn't work
But you do want rebase to run inside a same transaction if possible, so the
code may look like:
with repo.transaction():
....
if obsstore:
rebasemod.rebase(...)
obsstore.createmarkers(...)
if not obsstore:
rebasemod.rebase(...)
repair.strip(...)
That's ugly and error-prone. Ideally it's possible to just write:
with repo.transaction():
rebasemod.rebase(...)
saferemovenodes(...)
This patch is the first step towards that. It adds a "delayedstrip" method
to repair.py which maintains a postclose callback in the transaction object.
This is necessary because some callers in merge.py pass backgroundclose=True
when writing.
As with previous changes in this series, this should be a no-op.
We would like to migrate direct calls of repo.wvfs/wwrite/wread/etc to a
call on the relevant workingfilectx, both as a cleanup (to reduce the number of
working copy functions on `repo`), and also to facilitate an in-memory merge
that doesn't write to the working copy.
In order to do that, the first step is to ensure we pass the target wctx around
and perform our writes and reads on it. Later, this object might become a
memctx.
This is naive implementation using two-pass scanning. Tracking descendants
isn't an easy problem if both start and stop depths are specified. It's
impractical to remember all possible depths of each node while scanning from
roots to descendants because the number of depths explodes. Instead, we could
cache (min, max) depths as a good approximation and track ancestors back when
needed, but that's likely to have off-by-one bug.
Since this implementation appears not significantly slower, and is quite
straightforward, I think it's good enough for practical use cases. The time
and space complexity is O(n) ish.
revisions:
0) 1-pass scanning with (min, max)-depth cache (worst-case quadratic)
1) 2-pass scanning (this version)
repository:
mozilla-central
# descendants(0) (for reference)
*) 0.430353
# descendants(0, depth=1000)
0) 0.264889
1) 0.398289
# descendants(limit(tip:0, 1, offset=10000), depth=1000)
0) 0.025478
1) 0.029099
# descendants(0, depth=2000, startdepth=1000)
0) painfully slow (due to quadratic backtracking of ancestors)
1) 1.531138
Prepares for adding depth support. I want to process depth=0 in
revdescendants() to make things simpler.
only() also calls dagop.revdescendants(), but it filters out root revisions
explicitly. So this should cause no problem.
# descendants(0) using hg repo
0) 0.052380
1) 0.051226
# only(tip) using hg repo
0) 0.001433
1) 0.001425
Mercurial read all data between the base of the chain and the last delta when
restoring content (including unrelated delta). To monitor this, we add data
about the size of the "delta chain span" to debugrevlog.
Previously, rebase assumes the following pattern:
rebase:
with transaction as tr: # top-level
...
tr.__close__ writes rebasestate
unlink('rebasestate')
However it's possible that "rebase" was called inside a transaction:
with transaction as tr1:
rebase:
with transaction as tr2: # not top-level
...
tr2.__close__ does not write rebasestate
unlink('rebasestate')
tr1.__close__ writes rebasestate
That leaves a rebasestate on disk incorrectly.
This patch adds "removefilegenerator" to notify transaction code that the
state file is no longer needed therefore fixes the issue.
I meant to do this before sending the initial templater support, but forgot.
I'm quite surprised that 'dirty' doesn't occur in more user facing contexts, but
there are a few, like the help for blackbox. It also more obviously mirrors the
'(clean)' state printed by the summary command. I also didn't like that it was
just one letter off from {changes} in the {latesttags} sub-keywords, which has a
totally different meaning.
Before this patch, functions defined in extensions are registered via
extra loaders only in _dispatch(). Therefore, loading extensions in
other code paths like below omits registration of functions.
- WSGI service
- operation across repositories (e.g. subrepo)
- test-duplicateoptions.py, using extensions.loadall() directly
To register functions always at loading new extension, this patch
moves implementation for extra loading from dispatch._dispatch() to
extensions.loadall().
AFAIK, only commands module causes cyclic dependency between
extensions module, but this patch imports all related modules just
before extra loading in loadall(), in order to centralize them.
This patch makes extensions.py depend on many other modules, even
though extensions.py itself doesn't. It should be avoided if possible,
but I don't have any better idea. Some other places like below aren't
reasonable for extra loading, IMHO.
- specific function in newly added module:
existing callers of extensions.loadall() should invoke it, too
- hg.repository() or so:
no-repo commands aren't covered by this.
BTW, this patch removes _loaded.add(name) on relocation, because
dispatch._loaded is used only for extraloaders (for similar reason,
"exts" variable is removed, too).
This is based on a patch proposed last year by Mathias De Maré[1], with a few
changes.
- Tags and bookmarks are now formatted lists, for more flexible queries.
- The templater is populated whether or not [-nibtB] is specified. (Plain
output is unchanged.) This seems more consistent with other templated
commands.
- The 'id' property is a string, instead of a list.
- The parents of 'wdir()' have their own list of attributes.
I left 'id' as a string because it seems very useful for generating version
info. It's also a bit strange because the value and meaning changes depending
on whether or not --debug is passed (short vs full hash), whether the revision
is a merge or not (one hash or two, separated by a '+'), the working directory
or not (node vs p1node), and local or not (remote defaults to tip, and never has
'+'). The equivalent string built with {rev} seems much less useful, and I
couldn't think of a reasonable name, so I left it out.
The discussion seemed to be pointing towards having a list of nodes, with more
than one entry for a merge. It seems simpler to give the nodes a name, and use
{node} for the actual commit probed, especially now that there is a virtual node
for 'wdir()'.
Yuya mentioned using fm.nested() in that thread, so I did for the parent nodes.
I'm not sure if the plan is to fill in all of the context attributes in these
items, or if these nested items should simply be made {p1node} and {p1rev}.
I used ':' as the tag separator for consistency with {tags} in the log
templater. Likewise, bookmarks are separated by a space for consistency with
the corresponding log template.
[1] https://www.mercurial-scm.org/pipermail/mercurial-devel/2016-August/087039.html
This commit addresses a number of deficiencies in `hg show work`'s
output:
* Failure to render tags (it just wasn't implemented)
* Failure to render names associated with non-built-in namespaces
(e.g. remotenames)
* Color names were hardcoded instead of coming from the canonical
source in the namespace
This change has the intended effect of rendering tags and extra
namespaces. It solves an immediate need at Mozilla of having
names from a custom namespace printed, which is blocking us from
switching from a custom `hg wip` revset/template combo to `hg show
work`.
Note that the order of branches and bookmarks changes. This is
because bookmarks are registered before branches in namespaces.py.
We may want to register them last, after tags and branches. Or we
may want to added a weighted field to the namespace to control
display order. Something to think about.
I'm not a big fan of the complexity in the templating layer. There
is a lot of code to basically filter out the special case of
branch=='default' and tag=='tip'. Ideally, we would iterate over
a data structure that had irrelevant/unwanted names pre-filtered.
However, I wasn't sure how to best implement this. We probably
want {namespaces} to emit everything (its current behavior). I
was toying with the following:
* {namespacesnondefaults} variation that filtered values
* A filter function that operated on {namespaces} (I wasn't sure
how to implement this since the filtering layer would see a
"hybrid" instance as opposed to something that was definitely
an iterable of namespaces.)
* A namespaces(...) function where you could specify which values
to return. I like this the most. But it really wants named
arguments to control filtering and we only support named arguments
on revsets, not templates.
I figure perfect is the enemy of good and we can refine templating
support for namespaces in the future. At least now we have a
concrete example of a use case.
Currently, the templating layer tends to treat each namespace
as a one-off, with explicit usage of {bookmarks}, {tags}, {branch},
etc instead of using {namespaces}. It would be really useful if
we could iterate over namespaces and operate on them generically.
However, some consumers may wish to differentiate namespaces by
whether they are built-in to core Mercurial or provided by extensions.
Expected use cases include ignoring non-built-in namespaces or
emitting a generic label for non-built-in namespaces.
This commit introduces an attribute on namespace instances
that says whether the namespace is "built-in" and then exposes
this to the templating layer.
As part of this, we implement a reusable extension for defining
custom names on each changeset for testing. A second consumer
will be introduced in a subsequent commit.
Templates make use of a "log.<namespace>" label. The <namespace> value
here differs from the actual namespace name in that the namespace
itself is plural but the label/color value is singular.
Expose the color name to the templating layer so log.* labels
can be emitted for {namespaces}.
As part of this, we refactored the logic to eliminate a gnarly
comprehension. We store color names in their own dict because the
lookup can occur in tight loops and we shouldn't have to go to
repo.names[ns] multiple times for every changeset.
Previously, we constructed a formatter from a specific template
topic. Then from show() we reached into the internals of the
formatter to resolve a template string to be used to construct
a changeset templater.
A downside to this approach was it limited us to having the
entire template defined in a single entry in the map file. You
couldn't reference other entries in the map file and this would
lead to long templates and redundancy in the map file.
This commit teaches @showview how to instantiate a changeset
templater so we can construct a templater with full access to
the map file. To prove it works, we've split "showwork" into
components.
This just moves the BundleUnknownFeatureError exception handling one
level up so we collect the bundle2.applybundle{,1}() calls
together. applybundle1() will never throw the exception, so it should
have no functional consequence.
This is one step towards removing a bunch of "if isinstance(gen,
unbundle20)" by treating bundle1 and bundle2 more similarly.
The name may sounds ironic for a method in the bundle2 module, but I
didn't think it was worth it yet to create a new 'bundle' module that
depends on the 'bundle2' module. Besides, we'll inline the method
again later.
The results only need to be combined if they come from a bundle2. More
importantly, we'll change its argument to a bundleoperation soon, and
then it definitely will no longer belong in changegroup.py.
Before this patch, unbundling a stripped changeset would make it a
draft (unless the parent was secret). This meant that one would lose
phase information when stripping and unbundling secret changesets. The
same thing was true for public changesets. While stripping public
changesets is generally rare, it's done frequently by e.g. the
narrowhg extension.
We also include the phases in the temporary bundle, just in case
stripping were to fail after that point, so the user can still restore
the repo including phase information. Before this patch, the phases
were left untouched during the bundling and unbundling of the
temporary bundle. Only at the end of the transaction would
phasecache.filterunknown() be called to remove phase roots that were
no longer valid. We now need to call that also after the first
stripping, i.e. before applying the temporary bundle. Otherwise
unbundling the temporary bundle will cause a read of the phase cache
which has stripped changesets in the cache and that fails.
Like with obsmarkers, we unconditionally include the phases in the
bundle when stripping (when using bundle2, such as when generaldelta
is enabled). The reason for doing that for strip but not for bundle is
that strip bundles are not meant to be shared outside the repo, so we
don't care as much about compatibility.
This adds an experimental.bundle-phases config option to include phase
information in bundles. As with the recently added support for
bundling obsmarkers, the support for bundling phases is hidden behind
the config option until we decide to make a bundlespec v3 that
includes phases (and obsmarkers and ...).
We could perhaps use the listkeys format for this, but that's
considered obsolete according to Pierre-Yves. Instead, we introduce a
new "phase-heads" bundle part. The new part contains the phase heads
among the set of bundled revisions. It does not include those in
secret phase; any head in the bundle that is not mentioned in the
phase-heads part is assumed to be secret. As a special case, an empty
phase-heads part thus means that any changesets should be added in
secret phase. (If we ever add a fourth phase, we'll include secret in
the part and we'll add a version number.)
For now, phases are only included by "hg bundle", and not by
e.g. strip and rebase.
When adding support for bundling and unbundling phases, it will be
useful to have the list of added changesets. To do that, we return the
list from changegroup.apply().
Earlier I used pycompat.bytestr() to convert integers to bytes, but we can do
b"%d" % val to convert that int to bytes. b'' is already added by the
transformer.
Thanks to Yuya for suggesting this.
If an extension was loaded but disabled due to a minimumhgversion check it
will be present in the _extensions map, but set to None. The rest of the
extensions code treats the extension as if it were not present in this case,
but the afterloaded() function called the callback with loaded=True.
This allows even further customization via extensions for printing
bookmarks. For example, in hg-git this would allow printing remote refs
by just modifying the 'bmarks' parameter instead of reimplementing the
old commands.bookmarks method.
Furthermore, there is another benefit: now multiple extensions can
chain their custom data to bookmark printing. Previously, an extension
could have conflicting bookmark output due to which loaded first and
overrode commands.bookmarks. Now they can all play nicely together.
Note that this means that we're unnecessarily creating a transaction
in the pure "--inactive" (i.e. when deactivating the current
bookmark), but that should be harmless.
This is necessary to implement the set{gen} (set subscript) operator. For
example, set{-n} will be translated to ancestors(set, depth=n, startdepth=n).
https://www.mercurial-scm.org/wiki/RevsetOperatorPlan#ideas_from_mpm
The UI is undecided and I doubt if the startdepth option would be actually
useful, so the option is hidden for now. 'depth' could be extended to take
min:max range, in which case, integer depth should select a single generation.
ancestors(set, depth=:y) # scan up to y-th generation
ancestors(set, depth=x:) # skip until (x-1)-th generation
ancestors(set, depth=x) # select only x-th generation
Any ideas are welcomed.
# reverse(ancestors(tip)) using hg repo
3) 0.075951
4) 0.076175
Surprisingly, this makes revset benchmark slightly faster. I don't know why,
but it appears that wrapping -inputrev by tuple is the key. So I decided to
just enable depth computation by default.
# reverse(ancestors(tip)) using hg repo
1) 0.081051
2) 0.075408
Since we're using a max heap, the current rev should be a duplicate only
if it equals to the previous one. We don't have to maintain the whole seen
set.
# reverse(ancestors(tip)) using hg repo
0) 0.086420
1) 0.081051
- h -> pendingheap: "h" seems too short for variable of long lifetime
- current -> currev: future patches will add current "depth" variable
- parent -> prev or pctx: short lifetime, follows common naming rules
Add the followlines.js script and corresponding parameters as data attribute
on <tbody class="sourcelines"> element.
Extend CSS rules so that they also match the DOM structure of annotate view.
As previously, only address paper and gitweb styles (other styles do not have
followlines at all).
While plugging followlines.js into "annotate" view, we'll need to walk a
different DOM structure from that of "filerevision" view. In particular, the
selectable source line element is a <tr> in annotate view (in contrast with a
<span> in filerevision view). So make this tag name a parameter of
followlines.js script by passing its value as a "selectabletag" data attribute
of <pre class="sourcelines"> element.
As <pre class="sourcelines"> tags are getting quite long in templates, rewrite
them on several lines.
We will use this element to hook data attribute for the followlines.js script
to be plugged in annotate view. Also this gets symmetrical with paper style
which already has a <tbody> element.
c0593b622180 changed the styling of the "page_nav" CSS class to use
flexbox to separate elements within the <div>. I didn't realize that
this class was used outside of the links in the header. So this
resulted in incorrectly formatting links in the footer of various
pages. Fix that by introducing a new CSS class that preserves the
old CSS behavior.
If the option is registered, there is already a default value available and
passing a new one is at best redundant. So we issue a deprecation warning in
this case.
(note: there will be case were the default value will not be as simple as what
is currently possible. We'll upgrade the configitems code to handle them in
time.)
We do not have any registered config yet, but we are now ready to use them.
For now we ignore this feature for config access with "alternates". On the long
run, we expect alternates to be handled as "aliases" by the config item
themself.
The goal of this class is allow explicit declaration for the available config
option. This class will hold the data for one specific config item.
To keep it simple we start centralizing the handling of the default config value.
In the future we can expect more data to be carried on this class. For example:
- documentation,
- status (experimental, advanced, normal, deprecated),
- aliases,
- expected type,
- etc...
The old reversehunks code accesses "crecord.uihunk._hunk", which is the raw
recordhunk without crecord selection information, therefore "revert -i"
cannot revert individual lines, aka. issue5337.
The patch rewrites related logic to return the right reverse hunk for
revert. Namely,
1. "fromline" and "toline" are correctly swapped [1]
2. crecord.uihunk generates a correct reverse hunk [2]
Besides, reversehunks(hunks) will no longer modify its input "hunks", which
is more expected.
[1]: To explain why "fromline" and "toline" need to be swapped, take the
following example:
$ cat > a <<EOF
> 1
> 2
> 3
> 4
> EOF
$ cat > b <<EOF
> 2
> 3
> 5
> EOF
$ diff a b
1d0 <---- "1" is "fromline" and "0" is "toline"
< 1 and they are swapped if diff from the reversed direction
4c3 |
< 4 |
--- |
> 5 |
|
$ diff b a |
0a1 <---------+
> 1
3c4 <---- also "4c3" gets swapped to "3c4"
< 5
---
> 4
[2]: This is a bit tricky.
For example, given a file which is empty in working parent but has 3 lines
in working copy, and the user selection:
select hunk to discard
[x] +1
[ ] +2
[x] +3
The user intent is to drop "1" and "3" in working copy but keep "2", so the
reverse patch would be something like:
-1
2 (2 is a "context line")
-3
We cannot just take all selected lines and swap "-" and "+", which will be:
-1
-3
That patch won't apply because of "2". So the correct way is to insert "2"
as a "context line" by inserting it first then deleting it:
-2
+2
Therefore, the correct revert patch is:
-1
-2
+2
-3
It could be reordered to look more like a common diff hunk:
-1
-2
-3
+2
Note: It's possible to return multiple hunks so there won't be lines like
"-2", "+2". But the current implementation is much simpler.
For deletions, like the working parent has "1\n2\n3\n" and it was changed to
empty in working copy:
select hunk to discard
[x] -1
[ ] -2
[x] -3
The user intent is to drop the deletion of 1 and 3 (in other words, keep
those lines), but still delete "2".
The reverse patch is meant to be applied to working copy which is empty.
So the patch would be:
+1
+3
That is to say, there is no need to special handle the unselected "2" like
the above insertion case.
Changeset 3d003a7a1a87 change 'configwith' behavior so that the default value is
run through the conversion function. In parallel a new user of 'configwith' got
introduced unaware of this coming behavior change. This broke profiling.
We resolve the situation by having the new conversion function cope with a
default value already using the right type.
IIUC, letting the StopIteration through would not cause any bugs, but
not doing it makes the test-py3-commands.t pass.
I have also diligently gone through all uses of next() in our code
base. They either:
* are not called from a generator
* pass a default value to next()
* catch StopException
* work on infinite iterators
* request a fixed number of items that matches the generated number
* are about batching in wireproto which I didn't quite follow
I'd appreciate if Augie or someone else could take a look at the
wireproto batching and convince themselves that the next(batchable)
calls there will not raise a StopIteration.
These are the cases where either args is again passed as keyword argument or 1
or 2 elements are accessed. So it's better to add an r'' to prevent it
converting to bytes rather than doing the conversion of args.
This patch converts the args argument keys' to bytes wherever necessary as there
are some places where either args is not used or using r'' is better or args is
again passed as keyword arguments.
We were using str because on Python 2, str were bytes but now we have to use
bytes. Otherwise the if conditions fails and we have weird results from commands
on Python 3.
This should let 'configdate' delegate all special processing of the default
config value to the main 'config' method.
The default value for date (None) is still enforced in this method if no other
default were passed.
This should let 'configlist' delegate all special processing of the default
config value to the main 'config' method.
The default config value ([]) is still handled in this method.
This should let 'configwith' delegate all special processing of the default
config value to the main 'config' method.
This changeset introduce a small change in behavior since the default value is
run through the 'convert' function. This does not seems harmful and no actual
test break. This small change make the code simpler so I'm keeping it.
This should let 'configbool' delegate all special processing of the default
config value to the main 'config' method.
The default value for bool (False) is still enforced in this method if no other
default were passed.
We introduce a small object used to detect that no specific default value has
been passed to 'ui.config'. We need this explicit special value since "None" is
a valid and common default value.
The end goal here is to make progress on a centralised and explicit declaration
of the available config option. A first good usecase for this are "default"
value. Before starting looking further down this alley we needs to rework the
handling of default value in the 'ui' object to have all configxyz methods going
through the same logic. This is the first changeset on this trek.
By calling applybundle() with 'clonebundles' and the url instead of
calling processbundle(), the hooks will get different arguments:
HG_SOURCE will be 'clonebundles' instead of 'bundle2' and HG_URL will
be the url instead of 'bundle2'. This is consistent with the bundle1
behavior and seems like a bug fix, but I'm marking it BC anyway.
Again, commands.bookmark is getting too large. checkconflict already has
a lot of state and putting it in the bmstore makes more sense than
having it as a closure. This also allows extensions a place to override
this behavior.
While we're here, add a documentation string because, well, we should be
documenting more of our methods.
commands.bookmark has grown quite large with two closures already. Let's
split this up (and in the process allow extensions to override the
default behavior).
This can happen if another process (even another hg process!) comes along and
removes the file at that time.
This partly resolves issue5584, but not completely -- a bogus dirstate update
can still happen. However, the full fix is too involved for stable.
Since 1d07d9da84a0, pycompat.bytestr() wrapped by win32mbcs returns
unicode object, if an argument is not byte-str object. And this causes
unexpected failure at colorization.
pycompat.bytestr() is used to convert from color effect "int" value to
byte-str object in color module. Wrapped pycompat.bytestr() returns
unicode object for such "int" value, because it isn't byte-str.
If this returned unicode object is used to colorize non-ASCII byte-str
in cases below, UnicodeDecodeError is raised at an operation between
them.
- colorization uses "ansi" color mode, or
Even though this isn't default on Windows, user might use this
color mode for third party pager.
- ui.write() is buffered with labeled=True
Buffering causes "ansi" color mode internally, regardless of
actual color mode. With "win32" color mode, extra escape sequences
are omitted at writing data out.
For example, with "win32" color mode, "hg status" doesn't fail for
non-ASCII filenames, but "hg log" does for non-ASCII text, because
the latter implies buffered formatter.
There are many "color effect" value lines in color.py, and making them
byte-str objects isn't suitable for fixing on stable. In addition to
it, pycompat.bytestr will be used to get byte-str object from any
types other than int, too.
To resolve this issue, this patch does:
- replace pycompat.bytestr in checkwinfilename() with newly added
hook point util._filenamebytestr, and
- make win32mbcs reverse-wrap util._filenamebytestr
(this is a replacement of 1d07d9da84a0)
This patch does two things above at same time, because separately
applying the former change adds broken revision (from point of view of
win32mbcs) to stable branch.
"_" prefix is added to "filenamebytestr", because it is win32mbcs
specific hook point.
This commit essentially reverts 62ad9c1dbce9.
urllib2.URLError receives a "reason" argument. It isn't always a
tuple. Mozilla has experienced at least IndexError failures due
to the reason[1] access.
https://bugzilla.mozilla.org/show_bug.cgi?id=1364687
Before this patch, "hg CMD --pager on" on Windows shows output
unintentionally decorated with ANSI color escape sequences, if color
mode is "auto". This issue occurs in steps below.
1. dispatch() invokes ui.pager() at detection of "--pager on"
2. stdout of hg process is redirected into stdin of pager process
3. "ui.formatted" = True, because isatty(stdout) is so before (2)
4. color module is loaded for colorization
5. color.w32effects = None, because GetConsoleScreenBufferInfo()
fails on stdout redirected at (2)
6. "ansi" color mode is chosen, because of "not w32effects"
7. output is colorized in "ansi" mode because of "ui.formatted" = True
Even if "ansi" color mode is chosen, ordinarily redirected stdout
makes ui.formatted() return False, and colorization is avoided. But in
this issue case, "ui.formatted" = True at (3) forces output to be
colorized.
For correct console information on win32, it is needed to ensure that
color module is loaded before redirection of stdout for pagination.
BTW, if any of enabled extensions has "colortable" attribute, this
issue is avoided even before this patch, because color module is
imported as a part of loading such extension, and extension loading
occurs before setting up pager. For example, mq and keyword have
"colortable".
Adding markers to the repository might affect the set of obsolete changesets. So we
most remove the "volatile" set who rely in that data. We add two missing
invalidations after merging markers. This was caught by code change in the evolve
extensions tests.
This issues highlight that the current way to do things is a bit fragile,
however we keep things simple for stable.
Graft used findmissingrevs to find the candidates for graft duplicates in the
destination. That function operates with the constraint:
1. N is an ancestor of some node in 'heads'
2. N is not an ancestor of any node in 'common'
For our purpose, we do however have to work correctly in cases where the graft
set has multiple roots or where merges between graft ranges are skipped. The
only changesets we can be sure doesn't have ancestors that are grafts of any
changeset in the graftset, are the ones that are common ancestors of *all*
changesets in the graftset. We thus need:
2. N is not an ancestor of all nodes in 'common'
This change will graft more correctly, but it will also in some cases make
graft slower by making it search through a bigger and unnecessary large sets of
changes to find duplicates. In the general case of grafting individual or
linear sets, we do the same amount of work as before.
"merge-tools" help topic has described that the merge of the file
fails if no tool is found to merge binary or symlink, since
9da9bced2226 (or Mercurial 1.7), which based on (already removed)
MergeProgram wiki page.
But even at that revision, and of course now, merge of the file
doesn't fail automatically for binary/symlink. ":prompt" (or
equivalent logic) is used, if there is no appropriate tool
configuration for binary/symlink.
As an aside, I'm having trouble parsing the help text meaning for HG when it is
unset or empty. How can it be the frozen name or searched if it is empty?
This may be too subtle of a change to get the point across, but when I first
read the original text, I thought maybe the pager would only be invoked if
writing more than a screenful. The distinction between this and a pager that
simply exits after printing less than a screenful is important on Windows, given
the inability of `more` to color output.
Even though I figured this out a few weeks ago, I was initially puzzled where
the color went when I upgraded to 4.2 on a different Windows machine. Let's
point users reading the help into the right direction.
I wonder if we should be even more explicit about cmd.exe/MSYS/pager/color
interplay, but at least all of the breadcrumbs are here (I think).
This option was never released except for a release candidate. Dropping
compatibility with this option will free the 'pager.enable' config option for
other usage in the future.
This aligns with what we do for color (see cea7a760c58d). Pager is a central
enough notion that having the master config in the [ui] section makes senses. It
will helps with consistency, discoverability. It will also help having a simple
and clear example hgrc mentioning pager.
The previous form of the option had never been released in a non-rc version but
we keep it around for convenience. If both are set, 'ui.pager' take priority.
Previously, 'ui.color=yes' meant "always show color", While
"ui.color=auto" meant "use color automatically when it appears
sensible".
This feels problematic to some people because if an administrator has
disabled color with "ui.color=off", and a user turn it back on using
"color=on", it will get surprised (because it breaks their output when
redirected to a file.) This patch changes ui.color=true to only move the
default value of --color from "never" to "auto".
I'm not really in favor of this changes as I suspect the above case will
be pretty rare and I would rather keep the logic simpler. However, I'm
providing this patch to help the 4.2 release in the case were others
decide to make this changes.
Users that want to force colors without specifying --color on the
command line can use the 'ui.formatted' config knob, which had to be
enabled in a handful of tests for this patch.
Nice summary table (credit: Augie Fackler)
That is, before this patch:
+--------------------+--------------------+--------------------+
| | not a tty | a tty |
| | --color not set | --color not set |
| | | |
+--------------------+--------------------+--------------------+
| [ui] | | |
| color (not set) | no color | no color |
| | | |
+--------------------+--------------------+--------------------+
| [ui] | | |
| color = auto | no color | color |
| | | |
+--------------------+--------------------+--------------------+
| [ui] | | |
| color = yes | *color* | color |
| | | |
+--------------------+--------------------+--------------------+
| [ui] | | |
| color = no | no color | no color |
| | | |
+--------------------+--------------------+--------------------+
(if --color is specified, it always clobbers the setting in [ui])
and after this patch:
+--------------------+--------------------+--------------------+
| | not a tty | a tty |
| | --color not set | --color not set |
| | | |
+--------------------+--------------------+--------------------+
| [ui] | | |
| color (not set) | no color | no color |
| | | |
+--------------------+--------------------+--------------------+
| [ui] | | |
| color = auto | no color | color |
| | | |
+--------------------+--------------------+--------------------+
| [ui] | | |
| color = yes | *no color* | color |
| | | |
+--------------------+--------------------+--------------------+
| [ui] | | |
| color = no | no color | no color |
| | | |
+--------------------+--------------------+--------------------+
(if --color is specified, it always clobbers the setting in [ui])
"Churn" is not the useful example we have, but this is the one used in
'hg help config.extensions'. As we need something to replace the deprecated
'pager' extension in the example config, we are adding 'churn'.
Unlike a mapfile whose template is looked up by spec -> mapfile -> topic,
[templates] section is global. We use :sub-section syntax to define parts
per template.
As commented, this should be used with docheader and docfooter, not with
header nor footer. That's one reason why no props are passed to templater
when rendering a separator. (See map-cmdline.changelog to understand what
the "header" is.)
Since this postfix hack exists only for backward compatibility, we don't need
it for new [templates] section. This isn't a BC as templates defined in
[templates] section weren't loaded until recently.
changegroup.apply() currently creates a transation if there isn't
already one. Having the callers of that method pass in an existing
transaction seems a little cleaner. To do that, we need to make sure
all callers have a transaction. Since the transaction name is used as
a hook argument (HG_TXNNAME), we need to match the name from
changegroup.apply().
Several benefits:
* Gets close the comment describing it
* Splits off unrelated comment about "backup" argument
* Error checking is customarily done early
* If we added an early return to the method, it would still
consistently fail if there was an existing transaction (so
we would find and fix that case quickly)
One test needs updating with for this change, because we no longer
create the backup bundle before we fail. I don't see much reason to
create that backup bundle. If some command was adding content and then
trying to strip it as well within the transaction, we would have a
backup for the user, but the risk of that not being discovered in
development seems very small.
I have checked that all callers have already taken the lock (and if
they hadn't, we should have seen tests fail thanks to the 'transaction
requires locking' devel warning in localrepo.transaction()).
I'm not sure if this is better. If we're planning to add a template keyword
that returns obsoleted nodes unavailable in the repo (i.e. they have no valid
revision numbers), we might want to use the current "node"-only format
everywhere.
If a marker has no parent information, parents field is set to None, which
caused TypeError. I think this shouldn't normally happen, but somehow
I have such marker in my repo.
We point out at the help of the config option for user who wants to learn more
about the possible config value.
The suggested command returns some other extra (color related) results, but this
is bug to fix outside of the freeze.
The --color option is described to accept "boolean, always, auto,
never, or debug". Let's use a similar description for ui.color. Also
fix the formatting to use double quotes as we seem to use for about
half the values in config.txt (the other half uses double
backticks). Also use uppercase Boolean for consistency within the
file.
Acquiring lock by vfs.makelock() and getting lock holder (aka
"locker") information by vfs.readlock() aren't atomic action.
Therefore, failure of the former doesn't ensure success of the latter.
Before this patch, lock is unintentionally acquired, if ENOENT
causes failure of vfs.readlock() while 5 times retrying, because
lock._trylock() returns to caller silently after retrying, and
lock.lock() assumes that lock._trylock() returns successfully only if
lock is acquired.
In this case, lock symlink (or file) isn't created, even though lock
is treated as acquired in memory.
To avoid this issue, this patch makes lock._trylock() raise
LockHeld(EAGAIN) at the end of it, if lock isn't acquired while
retrying.
An empty "locker" meaning "busy for frequent lock/unlock by many
processes" might appear in an abortion message, if lock acquisition
fails. Therefore, this patch also does:
- use '%r' to increase visibility of "locker", even if it is empty
- show hint message to explain what empty "locker" means
Acquiring lock by vfs.makelock() and getting lock holder (aka
"locker") information by vfs.readlock() aren't atomic action.
Therefore, failure of the former doesn't ensure success of the latter.
Before this patch, lock is unintentionally acquired, if
self.parentlock is None (this is default value), and lock._readlock()
returns None for ENOENT at vfs.readlock(), because these None are
recognized as equal to each other.
In this case, lock symlink (or file) isn't created, even though lock
is treated as acquired in memory.
To avoid this issue, this patch retries lock acquisition immediately,
if lock._readlock() returns None "locker".
This issue will be covered by a test added in subsequent patch,
because simple emulation of ENOENT at vfs.readlock() easily causes
another issue. "raising ENOENT only at the first vfs.readlock()
invocation" is too complicated for unit test, IMHO.
Another raising PeerTransportError for "incomplete response" in
httppeer.py uses this (changed) hint message. This unification reduces
cost of translation.
There are some paragraphs, which aren't rendered in online help as
expected because of indentation and literal blocking issues.
- hgext/rebase.py
- paragraph before example code ends with ":", which treats
subsequent indented paragraphs as normal block
=> replace ":" with "::" to treat subsequent paragraphs as literal block
- help/pager.txt
- paragraph before a list of --pager option values ends with "::",
which treats subsequent indented paragraphs as literal block
=> replace "::" with ":" to treat subsequent paragraphs as normal block
- the second line of explanation for no/off --pager option value is
indented incorrectly (this also causes failure of "make" in doc)
=> indent correctly
- help/revisions.txt
- explanation following example code of "[revsetalias]" section
isn't suitable for literal block
=> un-indent explanation paragraph to treat it as normal block
- indentation of "For example" before example of tag() revset
predicate matching is meaningless
- descriptive text for tag() revset predicate matching isn't
suitable for literal block
=> un-indent concatenated two paragraphs to treat them as normal block
When on a filelog head, we are certain that there will be no descendant so the
target of the "descending" link will lead to an empty log result. Do not
display the link in this case.
We set parent._descendantrev = child.rev() when walking parents in
blockancestors() so that, when linkrev adjustment is perform for these, it
starts from a close descendant instead of possibly topmost introrev. (See
`self._adjustlinkrev(self._descendantrev)` in filectx._changeid().)
This is similar to changeset 8758896efb1c, which added a "f._changeid"
instruction in annotate() for the same purpose.
However, here, we set _descendantrev explicitly instead of relying on the
'_changeid' cached property being accessed (with effect to set _changeid
attribute) so that, in _parentfilectx() (called from parents()), we go through
`if '_changeid' in vars(self) [...]` branch in which instruction
`fctx._descendantrev = self.rev()` finally appears and does what we want.
With this, we can roughly get a 3x speedup (including in example of issue5538
from mozilla-central repository) on usage of followlines revset (and
equivalent hgweb request).
Update the hunk selector help message to use the operation name instead
of using "record" for all operations. Extract the help message in the same way
as other single and multiple message line.
Update tests to make sure that both "revert" and "discard" variants are tested.
Previously, calling blockancestors() with a fctx not touching file would
sometimes yield this filectx first, instead of the first "block ancestor",
because when compared to its parent it may have changes in specified line
range despite not touching the file at all.
Fixing this by starting the algorithm from the "base" filectx obtained using
fctx.introrev() (as done in annotate()).
In tests, add a changeset not touching file we want to follow lines of to
cover this case. Do this in test-annotate.t for followlines revset tests and
in test-hgweb-filelog.t for /log/<rev>/<file>?linerange=<from>:<to> tests.
Resolves test failures on FreeBSD, but I'm not happy about the fix.
A previous version of this also wrapped readline by putting the hack
in the _call method on doublepipe. That was confusing for readers and
wasn't necessary - just doing this on read() is sufficient to fix the
bugs I'm observing. We can always come back and do readline later if
needed.
`hg show work` is much more usable if output is colored. This patch
implements coloring via label() in a very hacky way.
In a default Mercurial install, you'll see yellow node labels for all
phases. Branches and bookmarks use the same formatting as the commit
message. So this change doesn't help much in a default install. But if
you have a custom colors defined for these things, output is much more
readable.
The implementation obviously needs some work. But for a minor change
on a feature that isn't convered by BC, this seems like a clear win
for the feature in 4.2.
Durham and I both like this better than "underway." We can add aliases
and bikeshed on the name during the 4.3 cycle, as this whole extension is
highly experimental.