Status of libarchive/bsdtar maintainership
Martin Matuska
mm at FreeBSD.org
Sat Feb 2 00:19:03 UTC 2019
I have created a pull request for the proposed patch. It breaks bsdtar's
test "test_missing_file" so I need to investigate.
https://github.com/libarchive/libarchive/pull/1131
On 02.02.19 00:35, Warner Losh wrote:
>
>
> On Thu, Jan 31, 2019 at 11:19 PM Eugene Grosbein <eugen at grosbein.net
> <mailto:eugen at grosbein.net>> wrote:
>
> On 01.02.2019 11:10, Warner Losh wrote:
>
> > On Thu, Jan 31, 2019, 8:22 PM Eugene Grosbein
> <eugen at grosbein.net <mailto:eugen at grosbein.net>
> <mailto:eugen at grosbein.net <mailto:eugen at grosbein.net>> wrote:
> >
> > Hi!
> >
> > I wonder what is status of our contrib/libarchive and
> bsdtar/bsdcpio etc. in modern versions of FreeBSD
> > in a sense of serious bug fixing. Long story short: I faced
> a bug in the libarchive bundled with 11.2
> > that makes it impossible to create reliable backups of live
> file system or its subtree
> > using cron+bsdtar utility that delegate actial work to the
> libarchive that just aborts
> > if a file disappears (is removed) in process (GNU tar
> continues with just warning).
> >
> > This is serious issue for me as I used 'tar' command to make
> backups for distinct subtrees
> > since FreeBSD 6.x and when my GPS+ntpd subsystem went insane
> and shifted system clock to 3 years
> > in the future, I lost data in several thousands of RRD
> databases and looked for backups to restore them
> > and found only small portion of databases in the tar instead
> of full backup.
> >
> > I've create the PR
> https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=233006 and later
> attached a patch
> > solving the problem in same way as GNU tar deals with it.
> >
> > Martin Matuska (mm) asked me to create an issue at GitHub
> for libarchive.
> > I have no GitHub account nor I need one, and he was so kind
> and created it himself:
> > https://github.com/libarchive/libarchive/issues/1082
> >
> > Almost 3 months have passed and no response from upstream.
> > Should we go ahead and fix it despite of it is part of contrib?
> >
> >
> > If you fix it, protocol is to submit it upstream first.
>
> That was done 3 months ago.
>
>
> I see the problem report in the github, but no pull request. Did I
> miss it?
>
>
> > It causes fewer problems in the long run. While it is tempting
> to just fix it in FreeBSD and move on,
> > almost every time we've done that in the past someone else has
> had to come in and fix the mess.
> >
> > Do you have a fix? Can you put it up for review somewhere?
>
> It is attached to mentioned PR:
> https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=233006#c6
>
>
> Did you submit it as a pull request? That seems to be how this
> upstream takes in code.
>
> > We are no where near a release, so there is no reason to rush
> this in.
>
> I waited for almost 3 months already. It seems, there would be no
> response at all.
>
>
> They didn't fix it in 3 months, sure. But it wasn't clear from the
> issue that you had an actual fix (I certainly missed that the first
> time through when I only looked at the github and not at our bug
> database). I'd try submitting a pull request and see what happens. I'd
> also send an email to mm@ telling him about the pull request and
> asking when he'll have time to look into integrating it or commenting
> on it. If he won't have time to get to it soon, I'd make the commit
> referencing the upstream pull request so the next person who imports
> things will notice if they tweak it before accepting the request.
>
> Warner
More information about the freebsd-stable
mailing list