dispatch: consolidate formatting of ParseErrors

error: store filename and message on LookupError for later It may be useful to be able to get to the failed path and the underlying error message when catching a LookupError, so let's make them accessible.

bundle2.unbundlepart: implement seek() This implements a seek() method for unbundlepart. This allows on-disk bundle2 parts to behave enough like files for bundlerepo to handle them. A future patch will add support for bundlerepo to read the bundle2 files that are written when the experimental.strip-bundle2-version config option is used.

bundle2.unbundlepart: tell() implementation Keep track of how many bytes we've returned from read(); fairly straightforward.

bundle2.unbundlepart: keep an index of chunks and their locations In order to make unbundlepart seekable, we need to keep a record of where the chunks are so that we can go back to the correct point.

bundle2.unbundlepart: raise payloadchunks from a closure to a method In a future patch, seek() will need to make a new chunk iterator for the stream; this places it somewhere it can be called multiple times.

manifest: make lru size configurable On machines with lots of ram, it's beneficial to increase the lru size of the manifest cache. On a large repo, configuring the lru to be size 10 can shave a large rebase (~12 commits) down from 95s to 70s.

parsers: use k instead of n for PyArg_ParseTuple because python 2.4 is awful

revset: have rev() validate input by repo.changelog.__contains__() fullreposet.__contains__() will be rewritten in order to support "null" revision, and "rev()" won't be possible to rely on it. This backs out eb763217152a, but there is no performance regression now. revisions: 0) e1dbe0b215ae "l not in fullreposet(repo)" 1) this patch "l not in repo.changelog" revset #0: rev(210000) 0) wall 0.000056 comb 0.000000 user 0.000000 sys 0.000000 (best of 48036) 1) wall 0.000049 comb 0.000000 user 0.000000 sys 0.000000 (best of 54969)

revlog: add __contains__ for fast membership test Because revlog implements __iter__, "rev in revlog" works but does silly O(n) lookup unexpectedly. So it seems good to add fast version of __contains__. This allows "rev in repo.changelog" in the next patch.

largefiles: set the extension as enabled locally after a clone requiring it When cloning a repo that requires largefiles, the user had to either enable the extension on the command line and then manually edit the local hgrc file after the clone, or just enable it globally for the user. Since it is a feature of last resort, and materially affects even repos without any largefiles when it is enabled, we should make it easier to not have it enabled globally. This simply adds the enabling statement to the local hgrc if the requires file mandates its use (which only happens after the first largefile is committed). That means that a user who works with a mix of largefile and normal repos can always clone with '--config extensions.largefiles=', and the extension is permanently enabled or not as appropriate. The change in test-largefiles.t is simply because the order of loading rebase and largefiles changed. The same change occurs if the order is flipped in the hgrc file at the top of the test.

color: be more conservative about setting ANSI mode on Windows (BC) The current color mode detection on Windows assumes the presence of the TERM environment variable assumes ANSI is supported. However, this isn't always true. In MSYS (commonly found as part of MinGW), TERM is set to "cygwin" and the auto resolved color mode of "ansi" results in escape sequences getting printed literally to the terminal. The output is very difficult to read and results in a bad user experience. A workaround is to activate the pager and have it attend all commands (GNU less in MSYS can render ANSI terminal sequences properly). In Cygwin, TERM is set to "xterm." Furthermore, Cygwin supports displaying these terminal sequences properly (unlike MSYS). This patch changes the mode auto-detection logic on Windows to be more conservative about selecting the "ansi" mode. We now only select the "ansi" mode if TERM is set and it contains the string "xterm" or if we were unable to talk to win32 APIs to determine the settings. There is a chance this may take away "ansi" from a terminal that actually supports it. The recourse for this would be to patch the detection to act appropriately and to override color.mode until that patch has landed. However, the author believes this regression is tolerable, since it means MSYS users won't have gibberish printed by default. Since MSYS's common pager (less) supports display of ANSI sequences, there is room to patch the color extensions so it can select the ANSI color mode if a pager is activated. Mozilla (being an active user of MSYS) would really appreciate this being part of the stable branch. However, since I believe it is BC, I haven't explicitly requested application to stable since I figure that request will be denied.

merge with stable

bundle2.unpackermixin: control for underlying file descriptor This patch adds seek(), tell(), and close() implementations for unpackermixin which forward to the file descriptor's implementation if possible. A future patch will use this to make bundle2.unbundlepart seekable, which will in turn make it usable as a file descriptor for bundlerepo.

trydiff: join elements in 'header' list by '\n' It seems natural that each element in the list corresponds to one line of output. That is currently true, but only because each element in the list has a trailing newline. Let's drop those newlines and instead add them when we print the headers.

trydiff: move check for quietness out of diffline() By moving the condition out of diffline(), the call site becomes clearer and diffline() no longer closes on any variables. Note that this changes the value of the header variable from [''] to [], but there is no difference in how these two are treated by the following code. The new value seems more natural anyway.

trydiff: remove dead branch in diffline() Since diffline() is never called when 'revs' is empty, it doesn't need to handle that case.

trydiff: make 'revs' ignored if opts.git is set Instead of setting revs=None to prevent the call to diffline() when opts.git is set, explicitly do not call the function in the git case.

trydiff: remove unused argument to diffline() Now that diffline no longer knows about copies/renames, it only needs one argument for the path.

trydiff: move git-header code out of diffline function This collects more of the code for writing git headers in a single place and makes diffline() close on a few variables less.

_fm1readmarkers: generate list in C This moves perfloadmarkers from ! result: 63866 ! wall 0.239217 comb 0.250000 user 0.240000 sys 0.010000 (best of 42) to ! result: 63866 ! wall 0.218795 comb 0.210000 user 0.210000 sys 0.000000 (best of 46)

obsolete: use parsers.fm1readmarker if it exists for a ~38% perf win This moves perfloadmarkers on my linux workstation (63494 markers from mpm, crew, and myself) performance from ! wall 0.357657 comb 0.360000 user 0.350000 sys 0.010000 (best of 28) to ! wall 0.222345 comb 0.220000 user 0.210000 sys 0.010000 (best of 41) which is a pretty good improvement. On my BSD machine, which is ancient and slow, before: ! wall 3.584964 comb 3.578125 user 3.539062 sys 0.039062 (best of 3) after: ! wall 2.267974 comb 2.265625 user 2.195312 sys 0.070312 (best of 5) I feel like we could do better by moving the whole generator function into C, but I didn't want to tackle that right away.

parsers: add fm1readmarker This lets us do most of the interesting work of parsing obsolete markers in C, which should provide significant time savings. Thanks to Martin von Zweigbergk for some cleanups on this code.

util: add getbefloat64 As far as I can tell, this is wrong. double's format isn't strictly specified in the C standard, but the wikipedia article implies that platforms implementing optional Annex F "IEC 60559 floating-point arithmetic" will work correctly. My local C experts believe doing *((double *) &t) is a strict aliasing violation, and that using a union is also one. Doing memcpy appears to be the least-undefined behavior possible.

util: add getbe{u,}int16 utility methods

obsolete: make optional offset parameter to fm*readmarkers required It was always passed by the only callsite, so just make it required.

log: fix json-formatted output when file copies are listed (issue4523)

merge with stable

copy: move _forwardcopies file logic to a function Moves the _forwardcopies missingfiles logic to a separate function so that other extensions which need to prefetch information about the files being processed have a hook point. This saves extensions from having to recompute this information themselves, and thus saves several seconds off of various commands (like rebase).

copy: move mergecopies file logic to a function Moves the mergecopies nonoverlap logic to a separate function so that other extensions which may need to prefetch information about the files being processed have a hook point. This saves extensions from having to recompute this information themselves, and thus saves several seconds off of various commands (like rebase).

histedit: allow configuring default behavior Adds a configuration setting for allowing users to specify the default behavior of 'hg histedit' without arguments. This saves users from having to manually figure out the bottom commit or a complicated revset. My current revset of choice is "only(.) & draft() - ::merge()" The commits that histedit can work with is usually quite limited, so if this feature ends up working well, we may want to consider making "only(.) & draft() - ::merge()" the default behavior for everyone.

revset: get revision number of each node from target namespaces Before this patch, revset predicate "named()" uses each nodes gotten from target namespaces directly. This causes problems below: - combination of other predicates doesn't work correctly, because they assume that revisions are listed up in number - "hg log" doesn't show any revisions for "named()" result, because: - "changeset_printer" stores formatted output for each revisions into dict with revision number (= ctx.rev()) as a key of them - "changeset_printer.flush(rev)" writes stored output for the specified revision, but - "commands.log" invokes it with the node, gotten from "named()" - "hg debugrevspec" shows nodes (= may be binary) directly Difference between revset predicate "tag()" and "named('tags')" in tests is fixed in subsequent patch.