Re: Cleaning up CVS tags from code

> > It just means it's recorded in the file so if we move the file from
> > that CVS to another one it keeps at least that information.
> 
> Not in my experience. Consider the $Id string in
> http://savannah.gnu.org/cgi-bin/viewcvs/classpathx/jaxp/source/gnu/xml/util/XMLWriter.java?rev=1.8&content-type=text/vnd.viewcvs-markup
> vs. the one in
> http://www.kaffe.org/cgi-bin/viewcvs.cgi/kaffe/libraries/javalib/gnu/xml/util/XMLWriter.java?rev=1.1&content-type=text/vnd.viewcvs-markup
> 
> it's completely different. the author, the date. just the file name is 
> the same. That's because the contents of those tags are automatically 
> insterted by RCS/CVS upon checkin, AFAIK. So when you move the file from 
>   one CVS to another, as in 'check it in', the old $Id tag gets 
> overwritten by a new one, causing a frivolous difference in files.

I didn't mean that one would do it without transformation on the
source file.

The scenario is that someone takes the tarball and puts it in a CVS
somewhere for some purpose. The $Id$ becomes a comment (after some
transform) and keeps the link to the original source.

 
> They seem to be purged from projects as soon as the projects become 
> important enough that people want to use those projects as 'upstream' 
> providers of sources. See discussions why those tags were purged from 
> GCC's Ada sources:
> 
> http://gcc.gnu.org/ml/gcc/2002-03/msg00557.html
> 
> here's a few more quotes from gcc developers:
> http://gcc.gnu.org/ml/gcc-patches/1999-08n/msg00760.html
> http://gcc.gnu.org/ml/gcc/2002-03/msg00560.html
> 
> See also this thread on lack of use of RCS tags in GNU emacs:
> http://mail.gnu.org/archive/html/bug-gnu-emacs/2001-05/msg00042.html
> 
> with GNU emacs developers' comments on evilness of tags:
> 
> http://mail.gnu.org/archive/html/bug-gnu-emacs/2001-05/msg00088.html
> http://mail.gnu.org/archive/html/bug-gnu-emacs/2001-05/msg00095.html
> http://mail.gnu.org/archive/html/bug-gnu-emacs/2001-05/msg00062.html
> 
> Just to show that I'm not making the problems up ;)
> 
> So I'm still interested in what RCS tags provide for you that cvs log 
> doesn't, or the cvsweb interface.

That's all interesting. I certainly wasn't aware of widespread GNU
dislike of them. However, I don't believe they're specifically
rejected by the coding standard.

How much of a problem are these for you? Why can't you simply
transform them out before you import the code?



Nic