Re: How well does CVS handle other types of data?

[Greg A. Woods]

[ On Saturday, July 14, 2001 at 14:34:42 (-0700), Mike Castle wrote: ]
> Subject: Re: How well does CVS handle other types of data?
>
> What bug?
> 
> CVS used to scan the contents of the conflict file to see if it had been
> fixed.  But that was changed to the current method.  Do you consider that a
> bug now?

Yes, absolutely.  It's a stupid unecessary bug too.

The change to make conflict marker checks only be a warning was made
without proper discussion and without review:

revision 1.111
date: 1997/09/14 22:24:07;  author: kingdon;  state: Exp;  lines: +12 -3
        * commit.c (check_fileproc): If the file has "conflict
        indicators", spit a warning and proceed with the checkin.
        * sanity.sh (conflicts): Adjust tests conflicts-132,
        conflicts-status-3, conflicts-133, and conflicts-status-4
        for new behavior.

Jim's rationale for it was purely selfish and not well thought out:

Jim Kingdon <address@hidden> wrote on <address@hidden>:
|
| It used to be firmer (error rather than warning).  But I got sick of
| having that code hit every time that sanity.sh got merged (and no, I
| don't consider your 4 command diff/patch wonder to be "pretty easy".
| No accounting for taste :-)).

Discussion on devel-cvs after he did this was split, and I don't think I
took part at the time, but from the looks of the archives nobody ever
provided any real examples outside of what turned out to be non-issues
in the CVS code itself.

Unforuntatly it seems the change was never backed out until consensus
could be reached and now people are treating it as if it's right thing
where it's clearly not!  Grrr.

-- 
                                                        Greg A. Woods

+1 416 218-0098      VE3TCP      <address@hidden>     <address@hidden>
Planix, Inc. <address@hidden>;   Secrets of the Weird <address@hidden>