Re: [gnome-love] another new face

[Elijah Newren <newren gmail com>]

On Thu, 9 Sep 2004 02:00:58 +0530, Baishampayan Ghose <b ghose gmail com> wrote:
> Don't CVS committers compile the source before they commit a patch
> into CVS? If they succeed in compiling on their local machine, why
> doesn't it compile later? And if it doesn't compile, why do they
> commit at al?

They _should_ do that.  But there's nothing that forces them to do so.
 And, of course, some people get lazy and commit anyway without taking
that step.

Also, just because it compiles on their machine, doesn't mean it will
compile on yours.  They may have forgotten to add a file to cvs, they
might only do a partial commit, they might make a change that depends
on a modification they made to another module (and which modification
they forgot to commit), there might be different versions of
automake/autoconf/freetype or whatever than you have, they might make
a change which depends on a change in a different module and they
committed both changes at the same time but you got the dependent
module checked out and built a few minutes before that checking, they
might run make without doing a make clean and forgetting that they
have a cached compile, etc. etc.

> > Were you compiling from CVS or from tarballs?  If from tarballs, you
> > have to understand that extra precautions are taken to make sure they
> > compile; tarballs don't change every minute of the day whereas CVS
> > HEAD can and sometimes does.

> It _was_ the CVS checkout which I got from somebody. Yeah, tarballs

And did they check it out at precisely the same time you did *and*
from the same cvs server?  If not, that could explain why the version
from them worked when your own version didn't.  In particular,
anonymous CVS is a day or so behind normal CVS and is only synced once
a day.  By getting a CVS checkout from someone else, you might have
been getting a version with an important bugfix to make something
compile.


Hope that helps explains some of the nasties that can occur,
Elijah