hgweb: just return iterables instead of yielding them

hgweb: separate out start_response() calling

hgweb: return iterable, add deprecation note

hgweb: remove some legacy code

keyword: avoid extra diffs when not comparing against working dir When diffing against specified revisions expanded keywords only get in the way. Update test output.

keyword: support mq; handle (q)record more gracefully mq: Ensure that expanded keywords do not make it into patches. - disable expansion when reading filelog - shrink expanded keywords when reading from working dir (wread) (q)record: Avoid additional hunks due to expanded keywords. However this is still a compromise, as keyword expansions are not updated in working directory because record should not overwrite files. Mention above shortcomings and "hg kwexpand" workaround in help and update test output. system argument parsing: Command detection might be slightly more expensive with dispatch._parse, but we will need this for improving "hg diff" output.

convert: enforce svn_source.getchanges() files unicity

localrepo: ensure files unicity in commit() (issue 714)

util: simplify unique

convert: fix invalid svn.ra.check_path() call (issue 771) The svn transport must be reparented to repository root before using check_path() with full entrypaths.

util_win32: make os_link more robust (issue 761) On mapped drives, os_link() manages to create links but nlink() does not report them.

transform a bunch of print statements to appropriate ui calls

Merge with crew

convert: document the subversion conversion model

convert: fetch less revisions when looking for a branch parent

Test subversion branches conversion

convert: make svn revision iterator interruptible

convert: fix cross-branches subversion revisions handling The "one revision belongs to one branch" assumptions is relaxed. Branch revisions are parsed down to the first branch copy encountered, older history is skipped. It means the conversion is still not satisfying when dealing with branches overwriting themselves. This issue already existed in the previous version.

convert: fix parents of last fetched svn revision

convert: don't scan directories on property changes

hooks: fix pre- and post- hooks specified in .hg/hgrc We were looking up hooks in the wrong ui object. Also, we weren't handling hooks to commands without a repo.

[PATCH] zsh_completion tag/mq fixes Been using these for almost a month now without problems, so I figured some other people might want them. This allows the zsh completion to work with tags and mq patch names containing spaces, and adds support for qgoto.

tests: Use /bin/sh for test-journal-exists, /bin/bash may not be available

patchbomb: prompt only once for SMTP password - simplify mail._sendmail to be a function rather than a class - simplify connect to return a function rather than a class - move exception handling from mail.sendmail to mail.connect - use a single connection for all messages in patchbomb

transactions: don't show a backtrace when journal exists Now we simply print a message on how to recover.

Merge with crew

diff: don't show function name by default We'd mistakenly made the -p option always on, which meant there was no way to turn it off. It also meant that we were sometimes splitting multibyte characters in function name, which isn't a good default.

merge with main

Merge with crew

convert should use default value when CVS_RSH is not set, that value being 'ssh'.

hgk: display branch name in diff header

Display debugstate dates as '%Y-%m-%d %H:%M:%S'

merge: add a bit more sanity to divergent copy checks

keyword: test that expansion is done filewise Checks issue fixed in 2a66138c5e7e.

Solaris compatibility fixes for test-keyword: - do not use /bin/false (yields 255 instead of 1) - do not use "cd -", simply use "cd .." instead.

convert: separate trunk detection from branch layout detection In some subversion repositories, trunk is present but no branches are used. The current code is assuming that both trunk and branches must exist before adding trunk's head to the heads list. It's just better to separate the branch layout stuff from the trunk one.

Merge with crew-stable

patch: handle empty vs no file in git patches (issue906)

patch: remove CRLF when parsing file names

Merge with crew-stable

Document --base null in bundle help

Merge with mpm

util_win32: implement posixfile_nt.writelines() Used by the keyword extension.

New extension to support problematic MBCS on Windows. The aim of this extension is to clear the problem related to having 0x5c in 2nd byte of encoded bytes. So this extension is usefull for: * Japanese Windows user shift_jis encoding. * Chinese Windows user using big5 encoding. To use this extension, simply enable it without any customization. Note that some important python built-in functions and mercurial functions are altered for this extension to convert argument if need to handle MBCS.

Fix not to use os.sep directly. This change is intended for MBCS support.

Add util.splitpath() and use it instead of using os.sep directly. This change is intended to allow hooking splitpath() by win32mbcs extension for MBCS support.

Add endswithsep() and use it instead of using os.sep and os.altsep directly. This change is intended to allow hooking endswithsep() by win32mbcs extension for MBCS support.

Use util.normpath() instead of direct path string operation. This change is intended to allow hooking normpath() by win32mbcs extension for MBCS support.

svn: fix copy exception

posixfile_nt: set closed early Exceptions in __init__ may leave us with close undefined in __del__

Don't use the -i option with sed This patch fixes test-keyword on platforms without GNU sed (like OpenBSD).

convert: respect hg.tagsbranch setting

Issue #880: Simple misspelling ('occured') in web error message.

Tweak finddate to pass date directly. Previous implementation was passing (date + "to" + date), which fails when "<" and friends are prepended to the date specifier. Signed-off-by: Mark Williamson <mark.williamson@cl.cam.ac.uk>

adding "prefix" option to "hg serve" (command line and [web] section) allows "hg serve" to serve from a different path than '/' (server root)

Fix XHTML validation error on file diff page. Replaces <div>, a block level element, with <span>, it's inline counterpart, in <pre>, which is also an inline element.

hook: redirect stdout to stderr for ssh and http servers

convert: add commandline.xargs(), use it in svn_sink class Introduce commandline.xargs() to limit argument list with respect to ARG_MAX bytes. If no ARG_MAX information available - use POSIX required minimum of 4096 bytes. Under Windows, while actual argument list length is limited to 32k, shells impose their own limits on command line length, down to 2047 bytes for cmd.exe under Windows NT/2k and about 2500 bytes for older 4nt.exe. See http://support.microsoft.com/kb/830473 for details about cmd.exe limitations. Since ARG_MAX is limit for argument list and environment, we reserve half of it and one byte for environment variables. This way with default ARG_MAX (4096 bytes) we get value 2047 bytes which is OK for Windows too.

keyword: fix some doc strings; update copyright

hg qrecord -- like record, but for mq I'm a former Darcs user, and I've discovered that it is very convenient to actually perform development using MQ first, and only when the patches are 'ready' move them to project's history in stone. Usually I work on some topic, temporarily forgetting about any version control, and just do coding, experimenting, debugging, etc. After some time, I approach a moment, where my work should actually go to patches/commits, and here is the problem:: As it is now, there is no way to put part of the changes into one patch, and another part of the changes into second patch. This works, but only when changes are touching separate files, and for semantically different changes touching the same file(s) there is now pretty way to put them into separate patches. For some time, I've tolerated the pain to run vim patches/... and move hunks between files by hand, but I think this affects my productivity badly. So, here is the first step towards untiing the problem: Let's use 'hg qrecord' for mq, like we use 'hg record' for usual commits!

cmdutil.commit: extract 'addremove' from opts carefully we are going to use cmdutil.commit for qrecord, and it's brother qrefresh does not support addremove.

dispatch: allow extensions to provide setup code we'll need this soon, when record extension will optionally depend on mq early -- when preparing cmdtable. Also, if accepted, ExtensionHowto wiki should be updated as well.

record: refactor record into generic record driver rationale --------- I'd like to make MQ version of record -- qrecord. >From the first glance it seemed to be easy -- the task in essence would be to change call to cmdutil.commit() to something like mq.qrefresh(). As it turned out queue.refresh() and cmdutil.commit() have different semantics -- cmdutil.commit() first scans for changes and then delegate the actual commit to lowlevel func. On the other hand queue.refresh() do it all in once, and I am a bit scary to change it. Maybe the right way would be to first refactor queue.refresh() to use cmdutil.commit() machinery, and then trivially adjust record, but I feel I'm not competent for the task right now. Instead, I propose we refactor record to be some sort of high-level driver, or like a high-level decorator one can say, which will first interactively filter changes, and then delegate commit job to high-level commiter, e.g. 'commit' or 'qrefresh' So, this patch does just that -- refactor record to be generic driver, and update 'hg record' code to use the driver. 'hg qrecord' will follow.

record: some docs While studing record, I've written some docstrings and comments. Hope they are useful.

keyword: clean up quiet setting in kwdemo and adding of untracked kwfiles

keyword: suppress keyword expansion for log commands With previous bugfix "hg log -p" expanded keywords. Turn off again by making log, tip nokwcommands.

keyword: (bugfix) only set changenode for kwtemplater when committing kwexpand must always obtain changenode from filectx, otherwise current changenode is expanded in every file. Also fixes "hg cat <more than 1 file>".

test-mq-header: don't fail with different patch(1) At least under FreeBSD patch(1) won't fail on empty patch and will print 'Hmm...' warning instead. Sanitize output so tests won't fail unexpectedly.

Fix up tests for XHTML validation bits

Fixed numerous XHTML problems that caused validation errors and warnings. Link elements for Atom and RSS feeds had no end tag. <div> block element inside <a> in-line element isn't right and caused rendering problems on Safari. Spurious </div> tag in search forms. Branch entry missing closing </a> tag.

patchbomb: attachment options changed '-a/--attach' send patches as attachment '-i/--inline' sends patches as inline attachment (old behavior of -a/--attach)

patchbomb: consistently use opts.get

patchbomb: fix more line continuations (coding style)

keyword: add test

Add extension for filewise RCS-keyword expansion in working dir - keywords are only expanded working dir, not in change history - completely customizable keywords/expansions based on hg templates - intrusiveness/slowdown can be reduced by specifying precise [keyword] file patterns - can be turned off/on anytime - changesets can be exchanged regardless of remote/local keyword settings

restore branch after rollback (issue 902)

context: fix filectx.parents() bug introduced when editing 180a3eee4b75

merge with crew-stable

Fix copies reporting in log and convert. If copy logged in file revision, we report copy for changeset only if file revisions linkrev points back to the changeset in question or both changeset parents contain different file revisions. This fixes extra copies reported when executable bit was changed for previously copied file.

context: preserve changeset in filectx if we have one If we know original changeset we are working with - try hard to preserve it. Fallback to filelog.linkrev() only if we have no way to get original changeset, since linkrev() may point to other changeset.

test-convert-hg-svn: make it more windows friendly

test-convert-svn-sink: wrap repetitive svn checks in a function

Merge with crew-stable

convert: Fix bug of limit_arglist() loosing file on limit boundary.

test-convert: test before() and after() conversion actions

test-convert-svn-source: test convert/multiple extra fields

Merge with crew-stable

util: drop params added during experimentation

hg tag: run tag hook just once

Make run-tests.py --interactive work on Windows

convert: filemap must call base converter before()/after() actions

fetch: hide authentication details

serve: Don't change directory in the child if invoked with -d and --cwd

Fix status char in color extension for deleted (missing) files.

merge with main

convert: print commit log message with local encoding correctly.

convert: Accept local path on win32.

color extension: change from GPL3 to 2

changelog: fix decoding of extra (spotted by pmezard)

convert: svn_sink: workaround of command line size limitation on win32. On win32, the command line arguments seems to be limited about 2400 bytes. So 'add', 'delete', 'propget' and 'propdel' operations are called several times within this limitation. And also 'commit' operation is changed to be called without passing target files not to exceeds this limitation because it cannot be a separated like other operations above.

convert: svn_sink: specify encoding of message file explicitly on commit.

mq: add --currentdate and --date options to qnew and qrefresh These options make qnew add and qrefresh update a "# Date "-style header line. This allows proper recording of creation / last modification dates of patches in patch queues. Note that `qrefresh -D` only updates existing header lines. It never adds them, and does not warn about this. This is because I expect people to have `[default] qrefresh -D` in their .hgrc so patches with tracked dates get updated, others are left unchanged. The suggested setup in .hgrc is, in fact, [default] qnew = -D -U qrefresh = -D I tried to not mix header styles, so `qnew -D -U` now writes the user in "# User "-style, while `qnew -U` still writes it "From: "-style. Also, if `qrefresh -U` must add the user, it does so in "# User "-style if the header contains a "# HG changeset patch" line. (This is caused by mq not supporting the "Date: "-style header line at all - a reasonable choice given its standard date format.)

Add colored output to status and qseries commands

patchbomb: simplify some line continuations

patchbomb: add linebreaks after colons (coding style)

Fixed XHTML validation errors

Move aliases section in help below synopsis (issue362) This should avoid confusion why e.g. "hg help co" shows help for "hg update" instead.

test-convert-svn-source: test stop revision

Sort commands table again and add some missing 'hg' in synopsises.

Obfuscate (instead of just escape) contact/owner in gitweb summary. This could contain an email address.

Don't let ui.username override web.contact (issue900) 4603eef60237 introduced using ui.username before web.contact, but this was never documented and might cause commit accidents. - Drop web.author (deprecated since 2005) - Try ui.username or $EMAIL as a fallback to display something useful. - Update docs for the fallbacks.

realign commands table

bisect: improve tests - test skipping - test noupdate - test bad->good search - test inconsistent state

bisect: handle search for bad to good transitions Automatically detect whether we're looking for a bad to good transition rather than the usual good to bad transition by detecting when badrev is inside the good set and flipping good/bad.

bisect: make bisect a built-in command

bisect: faster merging

bisect: merge ancestor lists when pushing to children - eliminate some redundant tests and assignments - move ancestor list merging to child update

bisect: skip calculations on candidates with too many ancestors Once an ancestor list has grown past the perfect threshold, all descendants are less optimal. Use a poison dict to avoid pointless operations on their long ancestor lists, thus eliminating most of the work.

bisect: stop early if we find a perfect candidate

bisect: find best node in ancestor collection pass

bisect: calculate candidate set while finding children

bisect: use a dict for children We fill in the children only for ancestors of badrev

bisect: propagate ancestor lists directly to children - calculate the children of all candidates - for each candidate, combine ancestor lists - pass ancestor lists to children - store ancestor count This eliminates the O(n**2) memory usage, while maintaining about the same performance.

bisect: add noupdate option

Merge backout

Backed out changeset c2a21fe60994 Unfortunately, this breaks backward compatibility.

Show number of changesets written to bundle files by default (issue569) This was only shown with -v before.

"C" status code considered harmful, so changed to "=" (issue899) C means conflict in other SCMs, and = is more intuitive anyway.

Adjust changed test output after 0145f9afb0e7 (trailing whitespace)

Removed tabs and trailing whitespace in python files

util: filter all st_mode with 0777 in checkexec

merge with crew

Clarify some messages during commit: - use "branch 'foo'" to distinguish from "branch merge". - commit messags can be empty (to abort commits) - Added value for editor message: Tell about HG: lines like CVS does.

merge with crew

Merge with crew-stable

Prompt for a commit message in editor, improve error message

patchbomb: make --bundle respect --desc

patchbomb: read bundle file in binary mode

prompt: kill matchflags Python already lets one to embed RE flags directly in a regex, which is a much nicer way to do things: (?iLmsux) (One or more letters from the set "i", "L", "m", "s", "u", "x".) ... matchflags was introduced in 67afecb8d6cc, and the record extension is the only user. I've killed matchflag, and adjusted record code appropriately.

test-convert-cvs: force cvs -f options to avoid cvsrc There are some reference to a CVS_OPTIONS environment variable doing that but it does not work on my system (cvs 1.11.20, macosx).

Patchbomb: Prompt password when using SMTP/TLS and no password in .hgrc. If .hgrc specifies SMTP/TLS and a username without specifying a password, prompt for a password.

highlight: fix more tracebacks by forcing util._encoding to hgweb.encoding This is needed in case util._encoding and hgweb.encoding conflict. Extreme example: HGENCODING=UTF-8 [web] encoding = ascii Note: display of files whose encoding differs from HGENCODING or hgweb.encoding behaves not exactly the same way as with highlight turned off as replacement chars are either converted or replaced.

hg tag: run tag hook just once

patchbomb: break lines > 80 chars (coding style)

changelog: inline trivial call for extra data unescaping

changelog: remove extract function

util: simplify unique

remove unused util.localsub function

add debugfsinfo to check fs capabilities

atomictempfile: avoid chmod weirdness on Linux vfat

checkexec: fix VFAT tempfile droppings with more modern Linux kernels More recent Linux kernels don't pretend to allow any bogus chmods on VFAT.

bisect: use proper locking when updating bisect.state

bisect: remove class - rename core function to _bisect - rename command function to bisect - move state management to bisect

bisect: keep history of all bad revisions - use a single state dict - find the minimum bad revision

bisect: use more standard command syntax and help

bisect: move reporting out of core bisect function

bisect: add skip command - read/write skip lines in state - skip candidates in skipnodes - move bisect begin logic to next - add skip command

bisect: simplify state handling and init - use a single bisect.state file - unify init and reset (delete the state) - move write into good/bad

bisect: remove unused imports

bisect: rename autobad/good/next

bisect: slightly improve the help message

bisect: remove stale test function

bisect: use array.array rather than lists for ancestor lists This nearly doubles performance and cuts memory usage in half on large bisections.

bisect: switch individual ancestor lists from dict to list This saves quite a lot of memory and increases performance

bisect: turn ancestors into an array This makes things faster and eliminates the separate stop hash

bisect: greatly simplify the ancestor accumulating loop

bisect: switch to rev-based calculation - use revlog.parentrevs in search loops - calculate stop directly in O(N) without reachable - move badrev check into candidates

bisect: fix up node vs rev naming

bisect: clarify some bisection code - rename __ancestors... to candidates - remove head parameter (it's always badrev) - eliminate comprehensions to shrink parents - num_ancestors -> n_child name fix - clarify node selection

bisect: inline num_children function

bisect: remove usage of sets

bisect: remove stop argument to ancestors function

bisect: use bail_if_changed - switch to using existing bail_if_changed code - only abort when we're about to do an update

cmdutil: make bail_if_changed bail on uncommitted merge

bisect: remove versionstr

bisect: simplify lookup - use repo.lookup directly - move merge check to check_clean

bisect: remove redundant clean checks

bisect: fold in two trivial functions

bisect: remove two unused functions

update test-merge-types

ui: allow default when prompting

merge: add flag merging intelligence - properly handle all 27 combinations of flags - prompt on conflicts

util: remove set_exec and set_link methods

patch: use util.set_flags

convert/svn: use util.set_flags

merge: use util.set_flags

wwrite: simplify with util.set_flags - always attempt to delete the file (gets rid of read-only message) - always write as a normal file - use set_flags to convert to link or set exec

util: add new set_flags method This is more efficient than calling set_link and set_exec

add --debug to test-merge-types

merge: simplify merge tests, fix exec flag bug If local changed exec bit and remote changed file contents, exec bit change would be lost.

hg: fail upon copy/link errors while cloning locally

convert: svn-sink: copy and set properties after adding dirs/files We can't store properties for files we haven't added to repo. Similarly, we can't copy file to directory we haven't added to svn yet. Remember needed changes and apply them in putcommit().

convert: don't read ~/.cvsrc in test-convert-cvs

merge with crew-stable

merge with crew-stable

convert: allow repo root to be matched in filemap For consistency with "rename subdir ." syntax in filemap, also support "rename . subdir".

.hgignore += tags & cscope files

Merge with crew-stable

merge with crew-stable

server: append to logfiles

Move diffing missing file test and remove test-notfound No point to have a test dedicated to this diff behaviour.

commands: add exits(1) if a specified file cannot be added (issue 891)

test-hgweb-commands: replace all "XX years ago" occurences

Switch to using cat with EOF doc instead of trying to quote newlines for echo.

Use skipped: instead of hghave: for skipping tests, use this in test-merge-types

merge with crew-stable

Use common readlink.py instead of own implementations per test script.

Use test -h instead of test -L for portability Citing from the GNU autoconf manual: "either form conforms to Posix 1003.1-2001, but older shells like Solaris 8 /bin/sh support only -h"

test-hgweb: fix css update

test-win32text: avoid unix2dos, printf extensions

Atom/RSS logo display problem using min font size. Using Minimum Font Size = 18 on Firefox, the Atom and RSS logos in a changeset web page look crushed together. No need to specify an absolute width for text; just use sensible padding instead.

man 5 hgrc implied you need to use external tools to do CRLF translation. This is no longer true; win32text provides a much better in-process alternative.

Merge with mpm

Merge with crew

Issue 882: add standard hook to reject text files with CRLF. While the win32text extension does LF <-> CRLF conversion, and will issue a warning in case a file already in the repository uses CRLF, it provides no mechanism for verifying that incoming changes use LF. In a large development team with some Windows users, it is virtually guaranteed that someone will forget to set up the encode filter correctly and accidentally check in a file using CRLF, which can cause warnings for other Windows users when they next fetch changes. Since this is a general problem it is desirable to have a pre-commit (or -push) hook available to reject such accidents earlier rather than trying to fix them up after the fact.

patchbomb: make "hg email -b" w/o destination work as advertised

mq: add --currentuser and --user options to qnew and qrefresh These options make qnew and qrefresh add/update the "From:" header (or, if present, the "# User" header). This allows proper attribution of patches in patch queues with multiple contributors.

Tests for b5605d88dc27 and 840e2b315c1f (interactive prompts during merge)

Make ui.prompt repeat on "unrecognized response" again (issue897) (this was accidentally changed in ca0d02222d6a)

Fix misleading error and prompts during update/merge (issue556)

patch: be more tolerant to diffstat failures (issue 865) Sometimes, revisions cannot be represented by a regular diff, only a git diff would capture binary files or permission changes. diffstat cannot handle git patches and will output "0 files changed" when fed with an empty diff. We cannot consider the latter to be an error, unless we rewrite diffstat to handle these correctly.

Fix chmod of writable but unowned files (issue530) This could happen e.g. in group writable local repositories where a file should become executable on update. (Patch by Benoit Boissinot attached to issue530)

Fix test-convert-darcs output on Debian etch (issue779) The CR characters of darcs get confused the output. Additionally this darcs version printed: darcs: <stdin>: hGetLine: end of file

fix spelling error

Merge with stable

Fix income/pull with bundle and -R (issue 820). Uses ui.setconfig() to tell bundlerepo where the main repo is. This is needed for when the --repository option is used. Adds tests to test-bundle and a new test script test-mq-pull-from-bundle, which plays out the situation that initially made me detect this bug (hg -R .hg/patches pull ../bundle.hg).

Add output markers to test-bundle so it's easier to spot the source of output diffs.

hgk: add black and blue as preferred colors

Remove spurious escape attempt in hg.1.txt

Use VISUAL in addition to EDITOR when choosing the editor to use.

util: get rid of is_win_9x wart

Add --verbose support to tags command.

Properly check tag's existence as a local/global tag when removing it.

[RFC] convert: fix --datesort The problem is that previously commit.date was used for sorting, but it's a string like "1 Jan xxx 2007", so it it wrong to use it for sorting. Another problem is that why we are using depth for sorting -- I have no clear answer -- it seems to be plain wrong. This patch is just an RFC.

highlight: pass encoding to lexers and formatter Try to avoid UnicodeDecodeError by: - encoding to local - passing util._encoding as lexer input encoding - passing hgweb.encoding as formatter output encoding

highlight: mandatory arguments where possible

patch: simplify directory creation

patch: write rej files for missing targets (issue 853)

patch: avoid file existence tests when possible in selectfile()

patch: move diff parsing in iterhunks generator