Comments on: Version control

By: Collaborative work with LaTeX « The Lumber Room

Collaborative work with LaTeX « The Lumber Room — Thu, 22 May 2008 15:20:14 +0000

[...] Here is a post by an actual mathematician using version [...]

By: kowalske

kowalske — Tue, 25 Mar 2008 18:13:45 +0000

I haven’t really tried to get co-authors to synchronize on version control, except for this book translation I mentioned. But then the access-right issue was simplified because the SVN server was on my home computer (accessible through dyndns), so I could set it up as desired. Indeed, I don’t think either Bordeaux then or ETH now would allow me to have an SVN server on the department computers.

I’m vaguely thinking of contacting the Origo team here to see if building a similar tool for (research)-paper-oriented projects would be feasible…

By: D. Eppstein

D. Eppstein — Tue, 25 Mar 2008 17:59:09 +0000

I similarly use cvs (for no better reason than what you give for using svn) for all my papers. The biggest problem arises when working with collaborators at other universities or even other departments at my own university: giving them access to my cvs repository is difficult, as the way I have it set up would require convincing my department’s computer support people to give them logins. In one case I’ve acquired a login at another university for exactly this reason; in others, I just exchange files by email and handle the version control part myself.

By: kowalske

kowalske — Mon, 24 Mar 2008 07:12:30 +0000

GIT (or distributed version control in general) does look like a good alternative. In fact, I started investigating it before I succeeded in getting SVN to work in Zürich. In the end, I decide to stay with it because it has worked well for me in terms of keeping my data safe, not because of the feature set itself (in fact, once there was a repository corruption due to hardware problems, and one of the SVN developers was kind enough to fix the database file himself, which is the kind of help one can often only dream of!). I have no reason to doubt that GIT will be as good, but one thing that seemed a bit shaky was the initial SVN-to-GIT import, to preserve history. There were two different scripts to do this, and the first one, though it claimed to have encountered no error, had only recovered a very partial view of the history. The second script (I don’t remember the names) apparently worked better, however.

By: Maurizio Monge

Maurizio Monge — Sun, 23 Mar 2008 19:16:26 +0000

Dear Emmanuel, i’m extremely happy to see that you decided to start a blog!
I think that using a VCS for a “latex project” is a very good idea, in case that you wanted to give a look at an alternative i recommend you to consider git, it is a distributed VCS and a repository is even easer to set up than svn (there is no difference between the local checkout and the remote repository, every checkout can be used as ‘source’), you keep locally a copy of the whole history, and it is very good when you have to work offline (you can commit to the local repository, and later push the updates to the ‘server’ or to repositories that you may have on another computer), and it encourages you to create branches to store possible alternative developments (that may eventually be merged into the main project).
If i will have some spare time, i’d really like to write a “git for Mathematicians Guide” some day.
Best regards, and thanks for sharing your experiences!