This recode call was removed in 597805361f86, because it looked the
encode(decode()) construct was a no-op. In fact, the first decode() call was
wrong, and entries still have to be encoded before being passed to the sink.
For some reason, if a copy source is deleted in the same revision it is
referenced, it is filtered out. This is silly, because this happens all the
time with move operations. Fortunately, the filtering code is buggy and ends
being a no-op 99% of the time, since it does not delete the right key. Just
remove all this nonsense.
* httprangereader: name, __iter__ and close are needed to mimic file object
* static-http opener:
- disallow write/append modes
- add (unused) atomictemp parameter
* static-http repo:
- root attribute is needed for localrepo.dirstate()
- _branch* attributes are required for commitctx and branchmap calls
* tags: force repo.opener.__iter__ call earlier to force httprangereader
to try to read the cache early, to avoid raising IOError later.
Previously, Mercurial assumed that the last word of the string
representation was the name of the moduled that was imported. This
assmption is incorrect, despite being true for the common case of an
exception raised by the Python VM.
For example, hgsubversion raises an ImportError with a helpful message
if the Subversion bindings were not found. The final word of this
message is not meaningful on its own, and is never the name of a
module.
This patch changes the output printed to be a simple stringification
of the exception instance. In most cases, this will be `abort: No
module named X!' rather than `abort: could not import module X!'.
No functionality change; all tests pass.
Recent Pythons (e.g. 2.6.5 and 3.1) introduce a change that causes
urlparse.urlunparse(urlparse.urlparse('x://')) to return 'x:' instead of 'x://'i and
urlparse.urlunparse(urlparse.urlparse('x:///y')) to return 'x:/y' instead of 'x:///y'.
Fix url.hidepassword() and url.removeauth() to handle these cases.
sys.stdout.write('-'*80 + '\n')
or
sys.stdout.write('-'*80 + '\r')
do not work on Windows as they do on unix. On a 80 columns Windows console, the
extra CR or LF are interpreted as if belonging to the next line, so the first
command displays 2 lines (only one on unix) and the second one leave the line
visible and move back to the following line. To avoid this, we sacrifice one
column under Windows.
For servers with branchmap support, the algorithm now works as follows:
1. A list of branches in outgoing changesets is created.
2. Using the remote branchmap, a check for new branches is performed.
3. A map (from branch to head list) of locally known remote heads before
the push is created, and one which, after step 4, will contain the locally
known remote heads after the push.
4. The post-push head map is updated with the outgoing changesets, using the
branch cache update mechanism.
5. A check for new heads is performed, by comparing the length of the head list
before and after push, for each branch. If there are new heads, an error
depending on whether or not there are incoming changes on the branch,
is returned.
6. If the push is allowed, a warning is written if there are incoming changes
on any branches involved in the push.
For old servers, an algorithm similar to step 4-6 above is used to check for
new topological heads only.
Two bugs are also fixed:
1. Sometimes you would be allowed to push new branch heads without --force.
A test for this case was added.
2. You would get the "note: unsynced remote changes!" warning if there were any
incoming changesets, even if they were on unrelated branches.
This bug happens if the filesystem doesn't support exec-bit, during merges,
for example in 0b01431fee25 on the hg repo.
If f is not in p1, but is in p2 and has the x-bit in p2, since the dirstate is
based on p1, and the FS doesn't support the exec-bit, the dirstate can't
"guess" the right bit.
We instead fix it in workingcontext.flags()/manifest.
Previously, it only returned revisions that were in the revlog when it
was originally opened; revisions added since then were invisible.
This broke revlog._partialmatch() and therefore repo.lookup().
(Credit to Benoit Boissinot for simplifying my original test script
and for the actual fix.)
This lets us change the threshold at which a *.d file will be split
out, which should make it much easier to construct test cases that
probe revlogs with a separate data file.
(issue2137)