43

I've just started using git for tracking changes in a LaTeX paper I'm writing, and for collaboration with co-authors.

However, I cannot find easily private git repositories with reasonable plans for scientific collaboration (I'm not asking about discounts, just about a different workflow, so no "one team of n developers", but there are "many small project, with different people").

There is ScribTeX, but unfortunately they are going down :/.

Other:

So, is there a (La)TeX-oriented Git (or, say, Mercurial) repository service?

That is, I'm looking for a repository (free or reasonably priced - i.e. for one with a PhD student salary, not a programmer salary :))

  • allowing for:
    • many collaborators,
    • many private repositories,
  • but can have strong limits on:
    • users per repository,
    • repository size.
9
  • 3
    You can install gitolite.
    – Paul Gaborit
    Commented Nov 12, 2012 at 23:24
  • 20
    Bitbucket offers unlimited private repos for academic users: atlassian.com/software/views/bitbucket-academic-license.jsp
    – Aditya
    Commented Nov 13, 2012 at 6:54
  • I confirm Aditya's comment, I'm currently using an academic licence for Bitbucket, and I have unlimited private repo for an unlimited number of collaborators.
    – user102
    Commented Nov 13, 2012 at 18:53
  • 2
    @seteropere I see, SugarSync is something like Dropbox - to backup and share folders. (I use the latter also for backing up research files, BTW.) However, it's not version control system indented for working on the same piece of code. It may fail when many people work on the same file. Plus, it gives to insight in changes, plus - it is not (as) easy to get any version you want. But to make it full + post screenshots - could you post is as a question (i.e. "Why use version control systems for writing a paper"). I would love to answer. Commented Nov 15, 2012 at 0:40
  • 1
    @PiotrMigdal Done it now. Sorry, I didn't realise the last editor wasn't the OP (confusing interface...). As a good practice, I try to have OP's consent before making changes that could "break the flow of the question" like this one. Commented Dec 24, 2015 at 8:58

6 Answers 6

36

Aditya's comment should be the accepted answer:

Bitbucket offers unlimited private repos for academic users.

3
  • Sounds the best, I will test it. Commented Nov 14, 2012 at 20:50
  • Is this for all academic personal or only students?
    – crsh
    Commented Nov 15, 2012 at 14:09
  • 2
    It seems to be for everyone. There are no specific terms, but they only ask What University are you associated with?, and test for an academic e-mail account. Commented Nov 15, 2012 at 16:44
18

Overleaf (used to be writelatex) now works with git. Latex is fantastic for typesetting academic papers properly, and Overleaf is great for writing latex collaboratively.

4
  • 2
    This is my standard solution for multi-institution LaTeX collaborations now, and the WYSIWYG interface of Overleaf means that even collaborators from normally non-LaTeX fields have been willing to work this way. One warning, though: Overleaf's backend appears to be proprietary and not actually git, and so many more sophisticated git tricks won't work with it. It also rejects file types that it doesn't believe are compatible with LaTeX.
    – jakebeal
    Commented Dec 23, 2015 at 15:22
  • 1
    @jakebeal: While people at Overleaf's open-source alternative ShareLaTeX continue to discuss on how to implement git integration or whether it's worth doing, in the first place, folks at closed-sourced Overleaf initiated relevant real-time collaborative editor open source project with git support. (to be continued) Commented Dec 24, 2015 at 9:39
  • 1
    @jakebeal: (cont'd) However, it seems that they realized the potential "danger" from this project to their commercial "big brother", as a brief look at JotGit's repository shows that the most recent change has been done in July 2014. BTW, those interested in truly, but feature-limited, multi-platform (that is, beyond only LaTeX) collaborative writing platform, might want to take a look at Authorea (see this post for more details). Commented Dec 24, 2015 at 9:40
  • 1
    I've been meaning to look into Authorea, but I haven't yet. We used Sharelatex for a while but Writelatex had better features so I reluctantly migrated despite liking the folks at Sharelatex & supporting their ideals. My tech decisions are driven by not just my own experience, but how my PhD students do as well. Commented Dec 26, 2015 at 19:54
14

Consider asking your research institute/university IT services.

  • In terms of fees, I'd consider such a tool as necessary for scientific work as your office chair, or backup disks/servers. My personal experience with asking for a git repository on a file server was that our director immediately answered "if that's what is needed for work, it needs to be installed." - that was it.

  • The far more important concern is, where research in progress and possibly data will end up. Storing such sensitive data outside the collaborating institutes is by default a big NO, even though lots of people send their research data by skype or dropbox.
    You'd be totally screwed up if you run into ownership/privacy problems.
    Even Bitbucket (who have "academic license") have write in their terms such statements (there are other statements about you retaining ownership of your data, privacy etc.):

    End User hereby grants Atlassian a non-exclusive license to copy, distribute, perform, display, store, modify, and otherwise use End User Data in connection with operating the Hosted Services.

  • For some areas of research it is even more NO, e.g. I work with patient data...

  • git works very well with distributed systems, even if they are only seldom connected. Worst case, people can email patches.

  • But installing git on a server is easy and people may get an ssh login, and that is all you need to have for your private git repository inside your institution IT structure, which avoids all that privacy trouble.

6
  • 2
    Good point with the privacy and tricky EULAs! Hopefully, in my case (theoretical physics) it is not an issue. But when it comes to asking local IT service... well, I have very mixed experience with them. That is, sometimes they can install you ask (and clearly not always), but also it's likely to break without a notice, in an unpredictable moment; I have had much, much better experience with external (usually commercial) services. Moreover, even in the most optimistically-unrealistic case (where everything works well), I won't be at the same place for more than a few years. Commented Nov 13, 2012 at 19:33
  • 3
    The restrictive words in that EULA are in connection with operating the Hosted Services. To me, it means that they may perform those operations if and only if they are related to providing you the Hosted Services, i.e., they can show you the source files when you click on your repositories or you write git pull, but they may not put them online on the internet or have their employees read them if it's a private repo. IANAL, but I'd trust them if this is the exact wording. Commented Nov 14, 2012 at 10:08
  • 1
    @PiotrMigdal: I agree with your IT services experience. One of the advantages of git is that no master is needed. You can just copy your repo and it will work. That is, 1) you can copy it and take it with you when you change jobs (if permitted by employer) 2) failure of the IT department's server does not affect syncing your local repo with the colleagues repos. Not convenient (ask for IP address, allow connection [talk to IT because of firewall - that hassle may make them repair the server ASAP], sync, deny connections again), but work can go on. You can also email patches.
    – cbeleites
    Commented Nov 14, 2012 at 11:51
  • @FedericoPoloni: Neither am I a lawyer. But I'd at least get official approval for using that service. As a (relevant) side note: the DFN (German national research and education network) has its own doodle replacement because of such concerns. And doodle is just about scheduling meetings, not about upload of your research.
    – cbeleites
    Commented Nov 14, 2012 at 12:12
  • 1
    @m.raynal: thank you. I may add that later on the git server got outside access for specified ssh keys (basically our own little VPN), so collaboration was possible with specific colleagues from other institutes. (It also gained a gitlab community edition web interface). For non-git-savy collegues, I nowadays use my own nextcloud server for collaboration: it includes a markdown editor, and I can of course have the files in a private git repo in addition.
    – cbeleites
    Commented Jan 21, 2020 at 13:40
7

I always found it difficult to teach people how to properly use repositories when collaborating on a manuscript. I have found www.authorea.com to work much better. Unfortunately, authorea suffers from the same problems you discussed in your question. Lack of private repositories.

2

If you really want to use Git. Then I recommend you install your own git server. There are two worthy clones of github: GitLab and Gogs.

Gogs Logo

Gogs - Go Git Service A painless self-hosted Git service

I prefer Gogs since it is just one file install based on GoLang and looks very pretty and has all the functionality you would expect: Code History, Issue Management, Wiki pages

1

If your collaborators aren't software developers themselves I wouldn't subject them to learning Latex.

It is not an easy task, believeme I've tried.

What you need is something like Google Docs that saves history of the document and allows you to collaborate.

Thankfully there is: https://www.sharelatex.com/

LaTeX, Evolved The easy to use, online, collaborative LaTeX edito

I'm not affiliated with them... just a happy user.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .