[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: conflicts-130 fails - more info
From: |
Derek R. Price |
Subject: |
Re: conflicts-130 fails - more info |
Date: |
Mon, 02 Apr 2001 13:15:34 -0400 |
Hi Pavel! Good to see you! :) Comments below.
Pavel Roskin wrote:
> > conflicts-130 fails on the today's CVS on RedHat 7.0 with ReiserFS (the
> > later may be suggestive, since it doesn't sort the files automatically).
> > However, the CVS from CVS passed "make check" few days ago on the same
> > machine.
>
> I have looked a bit deeper, this time with the CVS version as of "Mon Apr
> 2 15:45:08 UTC 2001". conflicts-130 fails if /tmp is on ramfs, but not on
> ReiserFS or ext2. I'm using Linux-2.4.2-ac28.
>
> As you can see from the listing below, "dir1" appears after "sdir" in the
> directory.
>
> $ ./sanity.sh -k ./cvs conflicts
> This test should produce no other output than this line, and a final "OK".
> FAIL: conflicts-130
> $ ls -alU /tmp/cvs-sanity/2/first-dir
> total 8
> drwxrwxr-x 1 proski proski 0 Apr 2 11:47 .
> drwxrwxr-x 1 proski proski 0 Apr 2 11:47 ..
> -rw-rw-r-- 1 proski proski 23 Apr 2 11:47 .#a.1.1
> drwxrwxr-x 1 proski proski 0 Apr 2 11:47 sdir
> drwxrwxr-x 1 proski proski 0 Apr 2 11:47 dir1
> -rw-rw-r-- 1 proski proski 64 Apr 2 11:47 a
> drwxrwxr-x 1 proski proski 0 Apr 2 11:47 CVS
> $
>
> I can "fix" this test by using dotest_sort, but then conflicts-136 fails.
> It appears that sanity.sh relies on the filesystem reporting filenames in
> a particular order.
>
> I want to get some feedback before I start fixing all the tests - it may
> take a lot of time.
>
> Another fix would be to require CVS to sort some lists. This time it's a
> list of unknown files.
Well, I checked in a patch for somebody recently which makes the VMS port
sort files since I was reluctant to change that many sanity.sh tests to use
dotest_sort, but I'm starting to believe that file order is none of CVS's
business (other than to maintain whatever order the file system returned),
with the possible exception of user specified file order and I'm not even
sure I want to go there.
Anyone think the answer isn't to sort the tests? We lose quite a bit of
information that way, but I think this may be the lesser of two evils.
On the other hand, it may be possible to write a dotest_sort which is smart
enough to sort in chunks and keep related info/warnings with the correct
filename. A bit of a chore, but it should be doable.
Derek
--
Derek Price CVS Solutions Architect ( http://CVSHome.org )
mailto:dprice@collab.net CollabNet ( http://collab.net )
--
If there be any among us who would wish to dissolve this Union or to change its
republican form, let them stand undisturbed as monuments of the safety with
which error of opinion may be tolerated where reason is left free to combat it.
- Thomas Jefferson; 1st Inaugural, 1801