Before, we didn't support lazy loading if absolute_import was in
effect and a fromlist was used. This meant that "from . import X"
wasn't lazy and performance could suffer as a result.
With this patch, we now support lazy loading for this scenario.
As part of developing this, I discovered issues when module names
are defined. Since the enforced import style only allows
"from X import Y" or "from .X import Y" in very few scenarios
when absolute_import is enabled - scenarios where Y is not a
module and thus there is nothing to lazy load - I decided to drop
support for this case instead of chasing down the errors. I don't
think much harm will come from this. But I'd like to take another
look once all modules are using absolute_import and I can see the
full extent of what is using names in absolute_import mode.
demandimport doesn't currently support absolute imports (level >= 0).
In preparation for this, we add some documentation and a code branch
to handle the absolute_import case.
The removed code was to support an __import__ function that doesn't
support the "level" argument. This argument was added in Python 2.5,
which we no longer support.
A few months ago, import-checker.py was taught to enforce a more
well-defined import style for files with absolute_import. However,
we stopped short of actually converting source files to use
absolute_import because of problems with certain files.
Investigation revealed the following problems with switching to
absolute_import universally:
1) import cycles result in import failure on Python 2.6
2) undetermined way to import C/pure modules
While these problems need to be solved, they can be put off.
This patch starts a series of converting files to absolute_import
that won't exhibit any of the aforementioned problems.
This means that users of request batching don't need to worry
themselves with capability checking. Instead, they can just use
batching, and if the remote server doesn't support batching for some
reason the wirepeer code will transparently un-batch the requests.
This will allow for some slight simplification in a handful of
places. Prior to this change, largefiles would have been silently
broken against a server which did not support batching.
"Real" batching only makes sense for wirepeers, but it greatly
simplifies the clients of peer instances if they can be ignorant to
actual batching capabilities of that peer. By moving the
not-really-batched batching code into peer.peer, all peer instances
now work with the batching API, thus simplifying users.
This leaves a couple of name forwards in wirepeer.py. Originally I had
planned to clean those up, but it kind of unclarifies other bits of
code that want to use batching, so I think it makes sense for the
names to stay exposed by wireproto. Specifically, almost nothing is
currently aware of peer (see largefiles.proto for an example), so
making them be aware of the peer module *and* the wireproto module
seems like some abstraction leakage. I *think* the right long-term fix
would actually be to make wireproto an implementation detail that
clients wouldn't need to know about, but I don't really know what that
would entail at the moment.
As far as I'm aware, no clients of batching in third-party extensions
will need updating, which is nice icing.
There is case where nodes are neither in tmpnodes nor leaf but still get
removed. For example, if you used the "edit" action, made a commit and run
--abort. The commit you made is not tracked by histedit, yet it will likely be
cleaned up with its parent. The commit may not tracked because no replacements
computations are done in the --abort case.
The for loop is already quite more complicated than necessary and we are about
to add some logic. Instead, we use a simple revset. Revset laziness should
provide us with similar performance.
The goal of this function is to strip content out of the repository. We do not
really care if this content is visible or cleanup node not and we should proceed
anyway. None of the internal actions are subject to this, however, a third party
extension running arbitrary commands during histedit is affected by this.
The conversion from git to hg was reading the remote branch list directly from
the origin server. If the origin's branch had moved forward since the last git
fetch, it would return a git hash which didn't exist locally, and therefore the
branch was not converted.
This changes it to rely on the local repo's refs/remotes list of branches
instead, so it's completely cut off from the server.
Now revset can parse nullary ":" operator and existing "foo+bar" tags, we
don't need the old-style parser.
This means scmutil.revsingle(), revpair() and revrange() no longer accept
a binary nodeid. An integer revision is still allowed as it isn't ambiguous.
Future patches will remove the old-style parser that happen to accept a
binary nodeid. A binary nodeid shouldn't be passed to scmutil.revrange()
because it is ambiguous. For example, bin('20' * 19 + '30') is valid binary
nodeid, but it can also be parsed as a revset expression, '0'.
The next patch will move the exceptional parsing of old-style ranges
to revset.tokenize(). This patch will allow us to see the result tree.
Note that the parsing result of '-a-b-c-' is incorrect at this changeset.
It will be fixed soon.
When highlight extension encountered files that pygments didn't recognize, it
used to fall back to text lexer. Also, pygments uses TextLexer for .txt files.
This lexer is noop by design.
On bigger files, however, doing the noop highlighting resulted in noticeable
extra CPU work and memory usage: to show a 1 MB text file, hgweb required about
0.7s more (on top of ~3.8s, Q8400) and consumed about 100 MB of RAM more (on
top of ~150 MB).
Let's just exit the function when it's clear that nothing will be highlighted.
Due to how this pygmentize function works (it modifies the template in-place),
we can just return from it and everything else will work as if highlight
extension wasn't enabled.
The process replacement is building a full mapping to allow moving bookmarks and
creating obsolescence marker. We do not need the full logic for abort so we
extract it. It will be useful as abort is missing some data about the
replacement and can crash when third party extensions push it a bit too far.
The faulty changeset use obsolescence marker to roll the repository back on
--abort. This is a problematic approach because --abort should be as close as an
actually transaction rollback as possible stripping all created data from the
repository (cf `hg rebase --abort` stripping all created changesets). Instead
3e883e7ec57b made all content created during the aborted histedit still
available in the repository adding obsolescence marker to make them hidden. This
will cause trouble to evolution user as a re-run of the same histedit (with
success) will likely result in the very same node to be "recreated" while
obsolescence marker would be in place for them. And canceling an obsoletion is
still a fairly complicated process.
This also rollback using obsmarkers instead of strip to clean up temporary node
on successful histedit run because the two change were not split in separated
changeset. Rolling that part back does not have significant consequence a will
have to be resubmitted independently
Previously, importing a case-only rename patch on a case insensitive filesystem
caused the original file to be marked as '!' in status. The source was being
forgotten properly in patch.workingbackend.close(), but the call it makes to
scmutil.marktouched() then put the file back into the 'n' state (but it was
still missing from the filesystem).
The cause of this was scmutil._interestingfiles() would walk dirstate,
and since dirstate was able to lstat() the old file via the new name,
was treating this as a forgotten file, not a removed file.
scmutil.marktouched() re-adds forgotten files, so dirstate got out of
sync with the filesystem.
This could be handled with less code in the "kind == regkind or kind
== lnkkind" branch of dirstate._walkexplicit(), but this avoids
filesystem accesses unless case collisions occur. _discoverpath() is
used instead of normalize(), since the dirstate case is given first
precedence, and the old file is still in it. What matters is the
actual case in the filesystem.
This check was a legacy bit from when the file data was being fetched manually
with 'ctx[wfn]', but archive() does that now. 2032c5b369f5 seems to indicate
that this avoided a problem where a merge adds a file to another branch, and
that test still passes.
Unfortunately, I don't see a way to create a test that modifies the file in the
temporary directory before the command exits.
I wonder if the os.lstat() call needs to be wrapped in an exception handler for
the case where archive didn't create a file because the file didn't exist in
that revision. But I wasn't able to trigger a problem without it on a real
repository.
Since 862fc5f60fef, .hgignore is ignored on Windows because a pat may have
a drive letter, but pathutil.join is posixpath.join.
"z:\foo\bar/z:\foo\bar\.hgignore"
Instead, this patch uses os.path.join() and util.localpath() to process both
parts as file-system paths.
Maybe we can remove os.path.join() at dirstate._ignore because 'include:' is
resolved relative to repo root? It was introduced by 29ce78e5f35c.
"Tags" and "Bookmarks" sections on this page already have the similar links,
and so does the "Branches" on summary page in gitweb, so let's do the same in
monoblue.
