Fix hg cat when the file cannot be found in the specified revision

Add support for cloning with hardlinks on windows. In order to use hardlinks, the win32file module is needed, and this is present in ActivePython. If it isn't present, or hardlinks are not supported on the underlying filesystem, a regular copy is used. When using hardlinks the biggest benefit is probably the saving in space, but cloning can be much quicker. For example cloning the Xen tree (non trivial) without an update goes from about 95s to 15s. Unix-like platforms should be unaffected, although should be more tolerant on filesystems that don't support hard links. (tweaked by mpm to deal with new copyfiles function) --- hg.orig/mercurial/commands.py 2005-09-13 19:32:53.000000000 -0500 +++ hg/mercurial/commands.py 2005-09-14 12:11:34.000000000 -0500 @@ -620,10 +620,6 @@ def clone(ui, source, dest=None, **opts) if other.dev() != -1: abspath = os.path.abspath(source) - copyfile = (os.stat(dest).st_dev == other.dev() - and getattr(os, 'link', None) or shutil.copy2) - if copyfile is not shutil.copy2: - ui.note("cloning by hardlink\n") # we use a lock here because if we race with commit, we can # end up with extra data in the cloned revlogs that's not @@ -638,7 +634,7 @@ def clone(ui, source, dest=None, **opts) for f in files.split(): src = os.path.join(source, ".hg", f) dst = os.path.join(dest, ".hg", f) - util.copyfiles(src, dst, copyfile) + util.copyfiles(src, dst) repo = hg.repository(ui, dest) Index: hg/mercurial/util.py =================================================================== --- hg.orig/mercurial/util.py 2005-09-08 00:15:25.000000000 -0500 +++ hg/mercurial/util.py 2005-09-14 12:16:49.000000000 -0500 @@ -12,7 +12,7 @@ platform-specific details from the core. import os, errno from demandload import * -demandload(globals(), "re cStringIO") +demandload(globals(), "re cStringIO shutil") def binary(s): """return true if a string is binary data using diff's heuristic""" @@ -217,17 +217,28 @@ def rename(src, dst): os.unlink(dst) os.rename(src, dst) -def copyfiles(src, dst, copyfile): - """Copy a directory tree, files are copied using 'copyfile'.""" +def copyfiles(src, dst, hardlink=None): + """Copy a directory tree using hardlinks if possible""" + + if hardlink is None: + hardlink = (os.stat(src).st_dev == + os.stat(os.path.dirname(dst)).st_dev) if os.path.isdir(src): os.mkdir(dst) for name in os.listdir(src): srcname = os.path.join(src, name) dstname = os.path.join(dst, name) - copyfiles(srcname, dstname, copyfile) + copyfiles(srcname, dstname, hardlink) else: - copyfile(src, dst) + if hardlink: + try: + os_link(src, dst) + except: + hardlink = False + shutil.copy2(src, dst) + else: + shutil.copy2(src, dst) def opener(base): """ @@ -244,13 +255,13 @@ def opener(base): if mode[0] != "r": try: - s = os.stat(f) + nlink = nlinks(f) except OSError: d = os.path.dirname(f) if not os.path.isdir(d): os.makedirs(d) else: - if s.st_nlink > 1: + if nlink > 1: file(f + ".tmp", "wb").write(file(f, "rb").read()) rename(f+".tmp", f) @@ -266,10 +277,41 @@ def _makelock_file(info, pathname): def _readlock_file(pathname): return file(pathname).read() +def nlinks(pathname): + """Return number of hardlinks for the given file.""" + return os.stat(pathname).st_nlink + +if hasattr(os, 'link'): + os_link = os.link +else: + def os_link(src, dst): + raise OSError(0, "Hardlinks not supported") + # Platform specific variants if os.name == 'nt': nulldev = 'NUL:' + try: # ActivePython can create hard links using win32file module + import win32file + + def os_link(src, dst): # NB will only succeed on NTFS + win32file.CreateHardLink(dst, src) + + def nlinks(pathname): + """Return number of hardlinks for the given file.""" + try: + fh = win32file.CreateFile(pathname, + win32file.GENERIC_READ, win32file.FILE_SHARE_READ, + None, win32file.OPEN_EXISTING, 0, None) + res = win32file.GetFileInformationByHandle(fh) + fh.Close() + return res[7] + except: + return os.stat(pathname).st_nlink + + except ImportError: + pass + def is_exec(f, last): return last

Convert hgk to use the hgit extension, and upate to the latest gitk

Turn hgit into an extension, and add commands supporting the latest gitk

Add log -b to show the branch a specific revision lives in This can be somewhat slow on very large repositories, so I didn't want to include it in -v --- hg.orig/mercurial/commands.py 2005-09-13 14:21:57.000000000 -0500 +++ hg/mercurial/commands.py 2005-09-13 18:55:52.000000000 -0500 @@ -1161,7 +1161,10 @@ def log(ui, repo, *pats, **opts): du = dui(ui) elif st == 'add': du.bump(rev) - show_changeset(du, repo, rev) + br = None + if opts['branch']: + br = repo.branchlookup([repo.changelog.node(rev)]) + show_changeset(du, repo, rev, brinfo=br) if opts['patch']: changenode = repo.changelog.node(rev) prev, other = repo.changelog.parents(changenode) @@ -1743,6 +1746,7 @@ table = { (log, [('I', 'include', [], 'include path in search'), ('X', 'exclude', [], 'exclude path from search'), + ('b', 'branch', None, 'show branches'), ('r', 'rev', [], 'revision'), ('p', 'patch', None, 'show patch')], 'hg log [-I] [-X] [-r REV]... [-p] [FILE]'),

Fix off by one in convert-repo tags --- crew.orig/contrib/convert-repo 2005-07-16 11:52:06.000000000 -0400 +++ crew/contrib/convert-repo 2005-09-02 02:58:14.000000000 -0400 @@ -86,9 +86,7 @@ class convert_git: for f in os.listdir(self.path + "/.git/refs/tags"): try: h = file(self.path + "/.git/refs/tags/" + f).read().strip() - p, a, d, m = self.getcommit(h) - if not p: p = [h] # git is ugly, don't blame me - tags[f] = p[0] + tags[f] = h except: pass return tags -- _______________________________________________ Mercurial mailing list Mercurial@selenic.com http://selenic.com/mailman/listinfo/mercurial

Fix bug with co -C across branches, update tests

Fix test-tag output for new abort message

Revert unrelated changes in previous commit

Fix abort message for clone

Revert some exception type changes in revlog

CONTRIBUTORS update

dirstate: two more stat -> lstat changes

Fix comment typo

Fix dangling symlink bug in dirstate walk code

Replace sequences of {ui.warn, return 1} with raise of util.Abort.

patchbomb: add TLS and SMTP AUTH support. Original patch from Lee Cantey <lcantey@gmail.com>.

Add --debugger global option With this option, hg will drop into the Python debugger on execution. Running 'continue' will execute normally, and the debugger will be reinvoked if an exception is raised.

Fix Windows status problem from new dirstate walk code

Update test-help for bundle/unbundle

Fix missing docstring for unbundle

clone: directory names take precedence over symbolic names If source matches both a directory name and a symbolic name to repo, the directory will be taken.

import: fix bug in mail-format detection

hgweb: fix scope for RepoError

Add preliminary support for the bundle and unbundle commands

Add missing import for hgwebdir

Cleanup of tabs and trailing spaces in documentation.

Add reporting instructions to unknown exception backtraces

Smarter handling of revlog key errors Use RevlogError for reporting exceptions Catch and report RevlogError exceptions at the command parser

Clean up hgweb imports Use demandloading Pull function-local imports up to demandload Scope hg and ui bits

Change grep -e to grep --all We want to reserve -e for future use and grep compatibility. I've changed every-match to all out of a general preference for shorter long option names where short options don't exist.

Fold import -m option into import -f

Revert silly TypeError change

Tighten up clone locking Add a lock on the destination, expand comment on source lock

Clean up local clone file list We now use an explicit list of files to copy during clone so that we don't copy anything we shouldn't.

Rewrite copytree as copyfiles This inverts the logic of copytree to allow copying single files at the top level.

Redo local clone hgrc fix We shouldn't copy hgrc on a local clone, nor localtags. Behavior should be identical to a remote clone.

Write out hgrc properly. Previously, we simply appended to the hgrc file, which meant that it ended up containing multiple "paths" sections. Now, we only modify "paths.default".

patchbomb: continue if we can't import readline.

Get test-commit to exit cleanly.

Commit date validation: more stringent checks, more useful error messages.

revlog: raise informative exception if file is missing.

Minor cleanups.

Created a class in util called chunkbuffer that buffers reads from an iterator over strings (aka chunks). Also added to util (for future use) is a generator function that iterates over a file n bytes at a time. Lastly, localrepo was changed to use this new chunkbuffer class when reading changegroups form the local repository.

Add test output.

Date validation must check for 32-bit width. Don't use assert to check.

Make date/timezone validation in changelog.add more robust. Add test.

Validate user input of dates when adding a changelog entry.

Add doc string for debugrename.

Fortify the recognition of a diff header.

Add -p to incoming and outgoing commands to show patch

Revert changeset 120aa5fc7ced1bf765b4f025f5a3a138cd87f49e. I inadvertantly used "hg import" on a message I saved, expecting it to do the right thing, but instead it made the commit look like I authored it, and filled the description with the email header. Changeset 737f9b90c571b3991f7e35dc846e7ef38a475cfe contains a guard against "hg import" doing this again.

Make import command reject patches that resemble email messages. See changeset 120aa5fc7ced1bf765b4f025f5a3a138cd87f49e for an example of why this is a good idea.

Make removal check more complete and informative.

Fix up remove command to use walk code.

From mercurial-bounces@selenic.com Thu Sep 1 07:01:32 2005 Return-Path: <mercurial-bounces@selenic.com> X-Original-To: bos@serpentine.com Delivered-To: bos@serpentine.com Received: from waste.org (waste.org [216.27.176.166]) by demesne.serpentine.com (Postfix) with ESMTP id 3616A20B571 for <bos@serpentine.com>; Thu, 1 Sep 2005 07:01:32 -0700 (PDT) Received: from waste.org (localhost [127.0.0.1]) by waste.org (8.13.4/8.13.4/Debian-3) with ESMTP id j81DxodQ028829; Thu, 1 Sep 2005 08:59:51 -0500 Received: from web32904.mail.mud.yahoo.com (web32904.mail.mud.yahoo.com [68.142.206.51]) by waste.org (8.13.4/8.13.4/Debian-3) with SMTP id j81DxnNA028824 for <mercurial@selenic.com>; Thu, 1 Sep 2005 08:59:49 -0500 Received: (qmail 25859 invoked by uid 60001); 1 Sep 2005 13:59:17 -0000 DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=Message-ID:Received:Date:From:Subject:To:In-Reply-To:MIME-Version:Content-Type:Content-Transfer-Encoding; b=O6sELrlCknW3M/gKVqijWs82e/CbDEum1sEitcuLKXaP9dHU175PszOqMgcSKykMY+BVXtcH3NeaXLM3FyBmqNkoPAvesezyFbgQsHSM1S028oOexybCKMvtGQJmz66hzd1fDb0QoPj1gCcGU2VDevQaOesSmo1xF9jJwy2LlLE= ; Message-ID: <20050901135917.25856.qmail@web32904.mail.mud.yahoo.com> Received: from [60.48.222.94] by web32904.mail.mud.yahoo.com via HTTP; Thu, 01 Sep 2005 06:59:17 PDT Date: Thu, 1 Sep 2005 06:59:17 -0700 (PDT) From: TK Soh <teekaysoh@yahoo.com> To: mercurial@selenic.com In-Reply-To: <20050828075808.GO27787@waste.org> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 X-Virus-Scanned: by amavisd-new Subject: Re: add -p to hg tip X-BeenThere: mercurial@selenic.com X-Mailman-Version: 2.1.5 Precedence: list List-Id: mercurial.selenic.com List-Unsubscribe: <http://selenic.com/mailman/listinfo/mercurial>, <mailto:mercurial-request@selenic.com?subject=unsubscribe> List-Archive: <http://www.selenic.com/pipermail/mercurial> List-Post: <mailto:mercurial@selenic.com> List-Help: <mailto:mercurial-request@selenic.com?subject=help> List-Subscribe: <http://selenic.com/mailman/listinfo/mercurial>, <mailto:mercurial-request@selenic.com?subject=subscribe> Sender: mercurial-bounces@selenic.com Errors-To: mercurial-bounces@selenic.com X-Spam-Checker-Version: SpamAssassin 3.0.4 (2005-06-05) on demesne.serpentine.com X-Spam-Level: X-Spam-Status: No, score=-1.6 required=5.0 tests=AWL,BAYES_00 autolearn=ham version=3.0.4 X-Evolution-Source: imap://bos@www.serpentine.com/ Content-Transfer-Encoding: 8bit

Merge with myself.

Output Content-encoding for tar.gz and tar.bz2 snapshots If the HTTP header 'Content-encoding' is missing for tar.gz snapshots, some gzip capable proxies can bork the tarball and serve unusable files to users. GZ tarballs are served using Content-encoding: gzip. BZ2 tarballs are served using Content-encoding: x-bzip2.

Fix up failing tests.

Optimize dirstate walking This generally cuts the time for hg status/diff in half, from 2s down to 1s. The main parts I'm trying to optimize are: 1) os.walk stats every file. dirstate.changes then stats every file again. 2) os.walk yields every file and subdir to dirstate.traverse who yields every file and everything in the dirstate map. dirstate.walk then filters this mass and yields every file to the caller. There should be fewer steps in here, and fewer duplicate strings yielded. 3) dirstate.walk runs util.unique on the results from dirstate.traverse, even though it is also passing things through dirstate.seen to look for duplicates. I've turned os.walk into something hg specific that takes all the dirstate ignore and matching rules into account. The new function also takes an function arg (statmatch()) the caller supplies to help filter out files it doesn't care about. dirstate.changes uses this to update state for each file, avoiding the second stat call. dirstate.walk is changed to turn the match function it is passed into a statmatch function. The only real difference is that a statmatch function takes the stat data as a second parameter. It now calls dirstate.walkhelper, who requires a statmatch function to be passed. This fails test-walk, but right now I think this is from a sorting error fixed by this patch. Index: crew/mercurial/dirstate.py ===================================================================