git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: Q: does index-pack work in place on windows?

From
Nicolas Pitre <nico@fluxnic.net>
Date
Feb 3, 2011, 05:32 UTC
Message-ID
<alpine.LFD.2.00.1102030026430.12104@xanadu.home>
In-Reply-To
<AANLkTi=m+kERSH7TJFjuRaUfA7FR03zwNn2PKFxrqiOp@mail.gmail.com>
On Wed, 2 Feb 2011, Erik Faye-Lund wrote:
Show 41 quoted lines
> On Wed, Feb 2, 2011 at 9:21 PM, Junio C Hamano <gitster@pobox.com> wrote:
> > ... or any other (operating / file) system where you cannot unlink a file
> > that is open?
> >
> > When you run
> >
> >    $ git clone git://some.where/repo/sitory.git local
> >    $ cd local
> >    $ git index-pack .git/objects/pack/pack-*.pack
> >
> > there is a call to write_idx_file() in builtin/index-pack.c, that feeds
> > the correct (and existing) name of the corresponding pack idx file.
> >
> > The callee in pack-write.c, after sorting the list of objects contained in
> > the pack, does this:
> >
> >        if (!index_name) {
> >                static char tmpfile[PATH_MAX];
> >                fd = odb_mkstemp(tmpfile, sizeof(tmpfile), "pack/tmp_idx_XXXXXX");
> >                index_name = xstrdup(tmpfile);
> >        } else {
> >                unlink(index_name);
> >                fd = open(index_name, O_CREAT|O_EXCL|O_WRONLY, 0600);
> >        }
> >
> > and then writes out the pack index to the file descriptor.  But index-pack
> > uses the usual has_sha1_file() and read_sha1_file() interface to validate
> > the "existing" objects, and is likely to have mmapped the .idx file when
> > it called use_pack_window().  Which makes me suspect that this unlink (or
> > the open that immediately follows) may fail on systems that do not allow
> > unlink on inode that has still users.
> 
> ...and you're right:
> 
> $ git index-pack
> .git/objects/pack/pack-d634271f4d7ca70c00148e967a343c3c46cd7705.pack
> Unlink of file '.git/objects/pack/pack-d634271f4d7ca70c00148e967a343c3c46cd7705.idx'
> failed. Should I try again? (y/n)? n
> fatal: unable to create
> '.git/objects/pack/pack-d634271f4d7ca70c00148e967a343c3c46cd7705.idx':
> File exists
Why would you do such thing in the first place?

If the pack exists along with its index on disk, there is no point recreating it. Maybe index-pack should just bail out early in that case.

Nicolas
Previous: Erik Faye-LundNext: Junio C Hamano
Message 3 of 7 in “Q: does index-pack work in place on windows?”
  1. Junio C HamanoFeb 2, 2011
  2. Erik Faye-LundFeb 2, 2011
  3. Nicolas PitreFeb 3, 2011
  4. Junio C HamanoFeb 3, 2011
  5. Nicolas PitreFeb 4, 2011
  6. Junio C HamanoFeb 4, 2011
  7. Nicolas PitreFeb 4, 2011

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.