Having read the replies, I thought I'd offer slightly different advice.
You are writing a dissertation. The formatting just needs to be what
satisfies your university's format requirements, which usually are
broad. I'd be surprised if they required an index for instance. Don't
waste time and effort on the formatting. For one thing, few people
will actually read your dissertation, unless what you're doing is
stupendous (and then they won't care about the format): your proof-reader
(you have got one, haven't you?), your supervisor, your examiners,
and ... that's usually about it. (Your parents will look at it.) If your
supervisor
supervisor can start fussing about the prettiness of (say) your equations
and tables rather than their content, you can reasonably suggest to him
that you
would appear to be finished. Just do a few test runs first of typical
equations
just to check that the output is at least reasonable.
Much later, when your topic turns out to be important again, someone like
me will remember seeing your dissertation mentioned, or find it through
Google^,
but I can assure you that by we'll still be more interested in the content.
I'd use the system with which you're most familiar. You don't want the
added distractions of trying to debug the typesetting software, and when
something goes wrong, it's much easier if you've used it before. (In my own
case, the night of the submission deadline, when I came to do
the final copy, I discovered that the troff installation Had Somehow Changed
and the output was completely messed up. Unfortunately that predated Plan 9
and yesterday(1),
but fortunately it's easy to check each stage of the pipeline, and
I could work out where to look for the change to undo.
If you're using troff, pick up a copy of refer from contrib.
soon I'll begin to write my thesis and I am planing to use ...