Re: Conflict committed

[David Wood]

Thank you for the advice about commitinfo. I believe we will do exactly 
that as a safeguard against future problems.

I totally respect that CVS should be content agnostic, although I did have 
to rename my "cvs" java package to "cvsutil" because you can't have dirs 
named "cvs" in the repository... 

It's too bad such a safeguard can't be offered as a kind of 
optional-but-default condition. Common case, etc.

address@hidden wrote on 10/29/2003 12:07:28 PM:

> The timestamp of the file being changed will let the file with conflicts
> be committed with just a warning.
> 
> Some folks consider it useful to check for conflict markers in their
> commitinfo scripts and refuse the commit.
> 
> It is not up to cvs to determine what data patterns should or should not
> be allowed to be committed into a file.
> 
>    -- Mark
> 
> Example:
> 
> % cd /tmp
> % mkdir testit
> % cd testit
> % cvs -d /tmp/testit/cvs-testrepos init
> % cvs -d /tmp/testit/cvs-testrepos co CVSROOT
> % cd CVSROOT/
> % (echo aaa; echo bbb; echo ccc) > file1
> % cvs add file1
> % cvs ci -mnew file1
> % echo ddd >> file1
> % cvs ci -madd file1
> % echo eee >> file1
> % cvs up -j1.2 -j1.1 file1
> % cat file1
> aaa
> bbb
> ccc
> <<<<<<< file1
> ddd
> eee
> =======
> >>>>>>> 1.1
> % touch file1
> % cvs ci -mconflict file1
> cvs commit: warning: file `file1' seems to still contain conflict 
indicators
> Checking in file1;
> /tmp/cvs-testrepos/CVSROOT/file1,v  <--  file1
> new revision: 1.3; previous revision: 1.2
> done
> cvs commit: Rebuilding administrative file database
> % 
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.2.3 (FreeBSD)
> 
> iD8DBQE/n/PQ3x41pRYZE/gRAqT1AKDVi92UQJZorCK47WNQcrSBtFPWOACg3T1T
> b6nolaHXENd1+/R0jqRIxZY=
> =k4nC
> -----END PGP SIGNATURE-----