ui: add configint function and tests

mq: strip extra whitespace from node ids in header (issue2790) This makes `hg qpush --exact` work with patches created by `hg export`.

check-code: disallow calling opener(...).read() and opener(..).write()

prevent transient leaks of file handle by using new helper functions These leaks may occur in environments that don't employ a reference counting GC, i.e. PyPy. This implies: - changing opener(...).read() calls to opener.read(...) - changing opener(...).write() calls to opener.write(...) - changing open(...).read(...) to util.readfile(...) - changing open(...).write(...) to util.writefile(...)

util & scmutil: adapt read/write helpers as request by mpm

store: use a subclass, not a function for fncacheopener

util.termwidth: never return 0 for terminal width Catch a case where the termwidth was being reported as 0 when I was connecting with TLMTask instead of NSTask in OSX. This caused the progress extension to print no progress. The termwidth should never return 0 so in case we would return 0, simply fall back to the default termwidth below which is 80.

discovery: add new set-based discovery Adds a new discovery method based on repeatedly sampling the still undecided subset of the local node graph to determine the set of nodes common to both the client and the server. For small differences between client and server, it uses about the same or slightly fewer roundtrips than the old tree-based discovery. For larger differences, it typically reduces the number of roundtrips drastically (from 150 to 4, for instance). The old discovery code now lives in treediscovery.py, the new code is in setdiscovery.py. Still missing is a hook for extensions to contribute nodes to the initial sample. For instance, Augie's remotebranches could contribute the last known state of the server's heads. Credits for the actual sampler and computing common heads instead of bases go to Benoit Boissinot.

debugbuilddag: use memctx for speed This drops the options to run arbitrary shell commands within commits and the option to create a file that gets appended to in every revision. It now supports to not write file data at all, which is very fast for generating a pure 00changelog.i (useful for discovery tests, for instance). Timings for 1000 linear nodes: * Old `hg debugbuilddag -o '+1000'` took 4.5 secs. * New `hg debugbuilddag -o '+1000'` takes 2 secs. * New `hg debugbuilddag '+1000'` takes 0.8 secs. (The last one creates only 00changelog.i).

localrepo: reuse parent manifest in commitctx if no files have changed This speeds up the in-memory version of debugbuilddag that I'm working on considerably for the case where we want to build just a 00changelog.i (for discovery tests, for instance). There are a couple of test changes because node ids in tests have changed. The changes to the patch names in test-mq-qdelete.t were required because they could collide with nodeid abbreviations and newly actually do (patch "c" collides with id "cafe..." for patch "b").

bundlerepo: fix and improve getremotechanges Fixes the regression where incoming could show local changes introduced by rev 72c84f24b420.

hgrc.5: mention new 'case collision' handling for ui.portablefilenames

win32: remove obsolete bookmarks and parentrevspec extensions from mercurial.ini

bundlerepo: drop struct

bisect: fix indent level

share: create 'hgrc' using an opener, like clone

backout 3a3584967a93 - the "unreachable code" was apparently not unreachable All tests repeatedly passes with 3a3584967a93 on some machines, but on other machines it regularly causes failure in test-mv-cp-st-diff.t, such as: @@ -203,6 +203,7 @@ - working to root: --rev 0 M a + M x/x A b a

merge with mpm

revset: avoid over-aggresive optimizations of non-filtering functions (issue2549) When limit, last, min and max were evaluated they worked on a reduced set in the wrong way. Now they work on an unrestricted set (the whole repo) and get limited later on.

convert: handle invalid subversion source paths

convert: also catch missing revlogs when introduced in repo roots The previous behaviour was almost as if convert.hg.ignoreerrors was always set for revisions without parents, except that errors were silently ignored. Revlog errors are handled as a side effect of getcopies(), but getcopies() was only called when convert.hg.ignoreerrors was set. Now we always call self.getcopies for root revisions, not only when convert.hg.ignoreerrors is set, just like we do on all other revisions. The extra call might be a bit expensive, but the proper fix for that would be to catch these errors in another way.

status: remove unreachable code for status between revisions The manifest value of a file will never be false when "not parentworking", and the expensive content comparision would thus fortunately never be reached. (If it was reached it would be wrong for example in case of renames.) This code once handled status against working directory, but that has been done elsewhere for a long time.

httprepo: proper handling of invalid responses without content-type (issue2019) This can currently be tested on http://sf.net/

httprepo: use the original exception after falling back to static-http failed Before the fine exception handling in httprepo was never shown. The static-http exception will now only be shown when static-http is requested explicitly.

test: test that backslash is preserved by the url class

tests: use raw string for url tests of '\' handling

color: fix using multiple effects

unbundler: separate delta and header parsing Add header parsing for changelog and manifest (currently no headers might change for next-gen bundle).

changegroup: new bundler API

bundlerepo: port to new bundle API

bundler: make parsechunk return the base revision of the delta

tests: add pyflakes checking for unused imports