One more vote for old trusted CVS here. Actually, I use OpenCVS since my OS of choice is OpenBSD.

I would like to bring to attention of SVN users the following package

http://www.ctan.org/tex-archive/macros/latex/contrib/svn-multi/

which allow you to create (TeX) log messages on the LaTeX document itself. There is another package for using SVN with LaTeX called svn. Disadvantage of svn is that is not playing well with the papers which are broken into multiple files. Here are couple very good articles from the PracTeX journal about SVN, LaTeX and even using it with TextMate

http://www.tug.org/pracjourn/2007-3/kalderon-svnmulti/

http://www.tug.org/pracjourn/2007-3/skiadas-svn/

http://www.tug.org/pracjourn/2007-3/scharrer/

Another interesting reading from the same issue of PracTeX Journal

http://www.tug.org/pracjourn/2007-3/henningsen/

which addresses original question.

I am not a big fun of distributed revision control systems like Git. If I would have to pick one distributed system it would be Mercurial.

I would like to point another issue. Traditional revision control systems are intended to be used with text files (source codes). Most classical revision control systems like CVS, SVN have troubles dealing with binary files. TeX files are of course text files. However, illustrations in papers unless created with the programming languages like PSTricks, MetaPost, PGF/TikZ, or Asymtote are binary files. So they are not plying well with the revision control systems. That is one more reason for young mathematicians to start programming their illustrations instead of creating them with vector graphics editor like Xfig. My favorite is PSTricks because it is essentially the easiest way to use pure PostScript inside TeX documents. As we all know PostScript is THE page description language. Therefore, I also have strong preference for Powerdot (PSTricks) over Beamer(PGF/TikZ).

Unfortunately, there is a disturbing trend at least when it comes to grant proposals of using proprietary binary formats like doc. Those formats do not play well with revision control systems which makes collaboration on grant writing very painful. Hopefully, as the "new" docx format which is default format of MS Office 2007 becomes more wide spread this problem will slowly disappear  (At least for the people that are familiar that docx is just bunch of XML files zipped together, only two of the files are actually important, one being XML version of document itself and another being Style Sheet file. People familiar with DocBook (apparently MS people were not one of them since they reinvented the wheel invented 1992) will know what I am talking about).

Cheers,

Predrag

One more vote for old trusted CVS here. Actually, I use OpenCVS since my OS of choice is OpenBSD.

I would like to bring to attention of SVN users the package svn-multi, which allows you to create (TeX) log messages on the LaTeX document itself. There is another package for using SVN with LaTeX called svn. Disadvantage of svn is that is not playing well with the papers which are broken into multiple files. Here are couple very good articles from the PracTeX journal about SVN, LaTeX and even using it with TextMate:

1. Kalderon - LaTeX and Subversion;

2. Skiadas, Kjosmoen, and Kalderon - Subversion and TextMate;

3. Scharrer - Version control of LaTeX documents with svn-multi.

Another interesting reading from the same issue of PracTeX Journal is Henningsen - Tools for collaborative writing of scientific LaTeX documents, which addresses the original question.

I am not a big fun of distributed revision control systems like Git. If I would have to pick one distributed system it would be Mercurial.

I would like to point out another issue. Traditional revision control systems are intended to be used with text files (source codes). Most classical revision control systems like CVS and SVN have troubles dealing with binary files. TeX files are of course text files. However, illustrations in papers, unless created with programming languages like PSTricks, MetaPost, PGF/TikZ, or Asymptote, are binary files. So they are not playing well with revision control system. That is one more reason for young mathematicians to start programming their illustrations instead of creating them with vector graphics editor like Xfig. My favorite is PSTricks because it is essentially the easiest way to use pure PostScript inside TeX documents. As we all know, PostScript is the page description language. Therefore, I also have strong preference for Powerdot (PSTricks) over Beamer  (PGF/TikZ).

Unfortunately, there is a disturbing trend at least when it comes to grant proposals of using proprietary binary formats like doc. Those formats do not play well with revision control systems which makes collaboration on grant writing very painful. Hopefully, as the "new" docx format, which is the default format of MS Office 2007, becomes more widespread, this problem will slowly disappear. (At least for the people that are familiar that docx is just bunch of XML files zipped together, only two of the files are actually important, one being an XML version of document itself and another being a Style Sheet file. People familiar with DocBook (apparently MS people were not one of them since they reinvented the wheel invented 1992) will know what I am talking about.)