RE: [Gnu-arch-users] Equivalent of CVS $Log$ and $Id$

[Parker, Ron]

> From: Matthieu Moy [mailto:address@hidden
> Sent: Wednesday, March 17, 2004 11:15 AM

> > Keeping versions within the text of a file is problematical,
> > for reason mentioned in the mailing list archives.

> http://article.gmane.org/gmane.comp.version-control.arch.user/21326

I'll add one to that list of reasons.  The problem is similar in nature to
the end-of-line-conversion issue that mucks up CVS, et al. or the
binary-vs-text issue that causes problems on any MS OS.  This requires that
the VCS is explicitly told which files are binary so that they don't get
corrupted.

I agree with the previous email that said this should be addressed in the
build process, rather than in the VCS.  Also I am not sure what the purpose
is having file level tags in a changeset based versioning system.  If what
is desired is, something along the lines of knowing which changeset last
altered a given file I might be able to understand that, but I don't see the
need to have it embedded in the file itself.  

On the other hand something along the lines of "cvs blame" or an annotate
function might be handy.  Right now I am doing some tla testing on Windows
when both of my workstations are compiling my actual work.  Once I am sure
everything works properly, I may look into coding these.  Right now I am
working through a tla tutorial script seeing what is still broken.
Unfortunately my personal machine is currently in the shop, so some of this
is on hold.  I don't know if what I am seeing is a Windows issue or a
tutorial issue.

-- 
Ron Parker
"First of all, do no harm...'--Hippocrates