Workflow for reviewing PDFs generated from TeX?

First of all, you are asking people to do you a favour. That's an important thing to remember and my answer would be different if this was about collaborators. Therefore, the number one principle is:

Make it as easy as possible for them.

Unfortunately, the exact answer is therefore going to be as variable as the people involved! But in case it's of any use, here are my recommendations (which can be broadly interpreted as being what I would like if I were being asked to do this):

1. Don't send anyone the whole thesis. Theses can get quite long and it's a bit daunting to be given the whole lot. People's effectiveness in proof-reading tends to slide a bit after a dozen pages (unless they're specially trained) so chop it up in to sections.

2. I really do find it much easier to proof-read on paper. I don't know what it is about it, but I spot far more when I'm reading it on paper rather than on screen. If you're only sending smallish sections, that's not a lot for any one person to print out. Plus you can make it so that it prints 2-up with reasonable margins but without shrinking the print size very much. Offer to pay for the printing cost, and then it's shouldn't be hard for a person to find a scanner to send you back the scanned copy - again, offer to pay for any costs (no-one will take you up on that).

3. For someone who does feel happy doing the proof-reading on-screen, I strongly recommend annotating the PDF. If the person has a graphics tablet then it's easy: get hold of one of the programs designed for annotating PDFs and let them use that. If you use jarnal then it's cross-platform (written in java) and it can also be installed on a central webserver so your friends and family don't have to install anything. Other free annotators are xournal and gournal for Linux. These can save their annotations as XML files or export them back as PDFs.

4. Give the person a one-page list of proof-readers' marks. An example can be found here, though it would be best to customise it a little. Most won't use it, but the idea is to make it clear that all you really need to know is where an error is, actually what the error is is less important as you'll be able to figure that out for yourself.

5. Provide a nice wide margin for comments. Add or subtract a lot from \hoffset to get the whole document as close to the edge as possible. Also, use line numbers so that big comments can be easily "anchored" to points in the text without long lines going this way and that.

6. To make it easy on yourself to integrate their comments, add in easily-findable tags that get shoved in to the margin every, say, paragraph. Then you can search for them rather than searching for words in the text. And if you do need to search for words in the text, make sure that your source is in one-sentence-per-line format as some programs aren't so good at searching when the desired text goes over a linebreak.

7. Lastly, don't send them back again. If you need it proof-read again, find some new friends. Ideally, proof-reading should be done at the very last stage and only when you've been through it so many times that you've started to think that "your's" is correct because you've seen it so often. Anyone not trained is going to be far less effective when seeing something for the second time. And you can never predict exactly what minor grammatical points the examiners are going to pick up on (mine was a rather over-heavy use of the word "then" - quite justified, I should add) so trying to be 100% perfect is senseless.

In case it's lost in that lot, let me repeat the names of those PDF annotators:

• jarnal. Java (so cross-platform), can also be used in client-server mode.
• xournal. C++ (so fast), Linux (AFAIK).
• gournal. Gtk2-Perl, Linux (AFAIK).

I've used both xournal and jarnal and found them great tools for just this purpose.

PDFSync is the way to go. Also, compile your thesis with the lineno package for line numbering (you can find it on CTAN with docs).

When anyone annotates your PDF and send it to you, PDFSync will allow you to just click on the passage in the PDF in order to open your LaTeX file in the corresponding position. It will save a lot of time when searching for missing commas and such things. For people that cannot annotate your PDF, they can refer to page and line number (thanks to lineno), and that will also greatly speed up your ability to find the place you need to do corrections, because you can just open your own PDF, and thanks to PDFSync you will be able to open the LaTeX source in the right place.

This is what I do now for my LaTeX writing, and it works wonders when passing review versions around to co-authors. Moreover, line numbers are used by APS when they send proofs for papers about to be published (not sure why they don't do it for referee versions too, but ...), and that makes the communication of typesetting errors much easier.

I can highly recommend Skim.app (BSD license) if they are using MacOSX. This tool has been valuable to my research and it provides various ways of commenting on pdf files.