temporary log file

[martin f krafft]

also sprach John A Meinel <john at arbash-meinel.com> [2006.02.11.1458 +0100]:
> I guess I don't see what is actually creating this file. Can you
> give us the full traceback from ~/.bzr.log?

The file is the temporary file to be used with $EDITOR. As others
have suspected, -m on the command line does not trigger the error.

Here's the traceback:

[24117] Mon 11:24:16.631 ERROR: exceptions.OSError: [Errno 13] Permission denied: './bzr_log.Hm8PIa'
  at /usr/lib/python2.4/tempfile.py line 236
  in _mkstemp_inner
Traceback (most recent call last):
  File "/usr/lib/python2.4/site-packages/bzrlib/commands.py", line 572, in run_bzr_catch_errors
    return run_bzr(argv)
  File "/usr/lib/python2.4/site-packages/bzrlib/commands.py", line 534, in run_bzr
    ret = cmd_obj.run_argv(argv)
  File "/usr/lib/python2.4/site-packages/bzrlib/commands.py", line 233, in run_argv
    return self.run(**all_cmd_args)
  File "/usr/lib/python2.4/site-packages/bzrlib/builtins.py", line 1299, in run
    message = edit_commit_message(template)
  File "/usr/lib/python2.4/site-packages/bzrlib/msgeditor.py", line 83, in edit_commit_message
    tmp_fileno, msgfilename = tempfile.mkstemp(prefix='bzr_log.', dir=u'.')
  File "/usr/lib/python2.4/tempfile.py", line 302, in mkstemp
    return _mkstemp_inner(dir, prefix, suffix, flags)
  File "/usr/lib/python2.4/tempfile.py", line 236, in _mkstemp_inner
    fd = _os.open(file, flags, 0600)
OSError: [Errno 13] Permission denied: './bzr_log.Hm8PIa'



also sprach Martin Pool <mbp at sourcefrog.net> [2006.02.13.0107 +0100]:
> Creating the log file in the working directory is also consistent
> with arch and baz.

Is that worth it? Do people really work with the log files at all,
or are they more an internal thing? Sure, it's a good idea to start
the log file on checkout and keep adding to it, but if you ask me,
there should be a command like `bzr add-log`, rather than to expect
people to work with ./bzr_log.R4Nd0M.

> One reason to put it there is that the working directory may be more
> protected against losing the user's valuable commit message -- for
> example some linux distros have /tmp in tmpfs (a ramdisk) or
> automatically erased at startup and so a power outage will lose their
> commit message.

Fair point. What about ./.bzr/log.R4Nd0M then?

> Perhaps it would be enough to make it in TEMPDIR and print the
> name if the commit is aborted?

How about we try . and only fall back to .bzr or TEMPDIR when the
first fails?

> > And I would to know why someone have read-only bzr versioned
> > directory and try to commit? Is it usual situation?
> 
> It does seem curious and unusual to me, but the original poster
> said he was controlling files in /etc.  It's reasonable to run bzr
> as a non-root user to record a history of files that that the user
> cannot themselves modify.

Our goal with this toy project is to make a root-server shared by
a few people who can configure the daily stuff without root access.
We want members of the group wheel to be able to add logcheck
filters. Obviously, we could set /etc/logcheck to 2770, but we have
it at 2750 simply because from logcheck's point of view, noone needs
to write the directory ever. I'd rather not loosen this restriction
because the versioning software does not support it.

-- 
 .''`.     martin f. krafft <madduck at debian.org>
: :'  :    proud Debian developer and author: http://debiansystem.info
`. `'`
  `-  Debian - when you have better things to do than fixing a system
 
Invalid/expired PGP (sub)keys? Use subkeys.pgp.net as keyserver!
 
"i have made good judgments in the past.
 i have made good judgments in the future." 
                                                      - george w. bush 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature (GPG/PGP)
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20060213/c4434c16/attachment.pgp