Documents in LaTeX, output in PDF, sources in SVN - need customizable component or software for peer review

We have a large SVN repo of LaTeX documents. Each document ultimately is rendered into PDF.

Documents need to be reviewed. Review has two major goals: ensure the quality of text itself and ensure the qualify of typesetting.

Right now reviewer could be开发者_Go百科 separated into two major groups:

1. Those who use SVN to checkout sources and build PDFs for themselves, and submit results of review as SVN commits

2. Those who get pre-built PDFs from ftp server and submit results as a free-form list of comments via email.

Document authors process the review results by processing \todo{} blocks in LaTeX code, by rolling back unneeded changes in SVN or incorporating free-form comments from e-mail into LaTeX sources.

Problem is, as the number of documents grow, it is very hard to keep track of the reviewers from the second group, and incorporate their suggestions in the timely and thorough manner.

Therefore, some nonsophisticated post-commit review solution is needed.

Requirements:

1. Post-commit review

2. All repository should be by default considered a review target without a need to specify files/line ranges for review.

3. Support for anonymous comments/review results

4. Reviews/comments should have status to facilitate processing

5. Web-based

6. Integration with SVN, so that new commits could automatically be incorporated in the review

7. Open source/cusomizable

ReviewBoard, CodeStiker, plugins for trac, Rietveld, JCR - they all fail to satisfy some of those requirements. Besides, most of them are too complex for the need at hand.

Crucible is nice. It is complex, but could be made easy via customization. However price tag is a bit heavy.

What did I miss?

UPD: we ended up without any automation after all. Power-users either have direct access to the repository and make reviews as branches/patches or as comments inside sources. Other people either store comments in plain-text, or annotate resulting PDFs.

After a while, we stopped looking for alternatives.

It seems like your requirements might be a little more complicated than you think if you've already looked at all those open source projects, and they all don't meet your requirements.

I agree with using a DVCS if you can. I use git to manage all my latex code, and it's awesome.

have you thought of using just a plain text file to store your reviews? You could use something like org-mode in emacs, set up relative links to your documents, time stamp your reviews, use the TODO status cycling to keep track of what's reviewed and whats not, and keep the reviews right along side the latex code, something like: document.tex and document_review.org,.txt or whatever. org-mode will publish to latex, html, and a few other formats.

Anyway, just a suggestion-- KISS principle at work!

This sounds like an interesting workflow. Just how many documents are you considering, and how do they get into the system? It sounds like a government client...

As a first recommendation, EasyChair. I've had good experiences with it, in managing the process from author submission, through review, to the selecting accepted papers by the program committee. It's written in Perl, and I have heard it is reasonable easy to work with the sources, although I haven't contact with that. The handling of anonymising, in particular, is well done. — Not open source. I know one of the developers, and maybe the code can be shared on some basis, but I guess that rules this out.

EasyChair might be able to be integrated with your version control system. To be honest, I don't see why this is on your list. You want referees to work with one particular version of the text. It is with the copy-editor where the document will pass through several versions, and you can expect the contact to be less formal there.

Are you really wedded to Subversion? If there is substantial back-and-forth between authors and copy-editors (as opposed to the formalised relationship between authors and referees), having true distributed version control with good quality three-way diffs makes it much easier to have both authors and editors make changes to the text as they go along. I know of some publishers who use Git Hub to handle Latex submissions for this reason.