Some time ago there was this question from Igor on biblatex and REVTeX. Pretty much everyone in Physics is locked into using REVTeX because some of the most important journals require it, and it's stopping us from migrating to biblatex. Given this problem, I've been pondering writing a question on how to hack together a solution. There is a good chance this means changing a lot of code in the class file, but honestly I'm not sure. I've never had to dig into this kind of thing before. It could even be an entirely new macro package.

So the question is, will this be too big, even with a large bounty? At what point does a problem become too large to reasonably ask TeX.sx?


I have accepted an answer, although at the time of writing this note the answers by Joseph Wright, lockstep and Andrew Stacey all raised good points, and are worth a read.

  • 2
    I suspect if you ask the question my answer will be 'Do not do this: you do not know how much the publishers back-end is tied to the REVTeX internals.' – Joseph Wright Nov 27 '11 at 10:31
  • @JosephWright: Well, they always ask the content of the generated bbl file to be included, so they never run BibTeX on their end anyway... – qubyte Nov 27 '11 at 10:51
  • Well yes, but the .bbl file you get from biblatex is very different from that using natbib. I had some discussions with a publisher recently about their TeX support, and in their BibTeX style there is some XML included in the .bbl for the publisher. So any deviation from the official route would be a serious problem. – Joseph Wright Nov 27 '11 at 11:08
  • @JosephWright: Ah, I see now. Nonetheless the final part of the question stands. When does a question get too big? – qubyte Nov 27 '11 at 11:12

'How big is too big' is a difficult one, because at least in part it depends on the person answering. For 'big' problems, I suspect that at least on TeX.sx the size of the bounty you offer is not the key driver. It's more about the question, and whether someone wants to take it up. I think that asking a question that specifically requires writing a major package is probably 'too big', but asking a question that is simply difficult cannot be.

What I do think is important is to consider whether a big problem can be broken down. For example, in the case you raise you could ask about altering publishers class files in general, and whether there are any obvious pitfalls. You could also ask a more general question about loading biblatex with a class that has already loaded natbib (REVTeX doesn't just load natbib, it also has a significant amount of internal code that requires natbib.)

  • I somewhat disagree what the "major package" argument, but +1 for the hint to break down big problems. – lockstep Nov 27 '11 at 11:22
  • @lockstep I guess I'm saying that asking for a major package as the key part of a single question probably means it should be broken down. – Joseph Wright Nov 27 '11 at 11:24
  • @lockstep: Breaking the question down is a good idea, but I struggle to find ways in which to do that for this problem. Suggestions? – qubyte Nov 27 '11 at 14:25
  • @MarkS.Everitt: No (as I don't know anything about REVTeX). – lockstep Nov 27 '11 at 14:46
  • @lockstep: Lucky for you! ;) – qubyte Nov 27 '11 at 14:48
  • 2
    @MarkS.Everitt As I said, one 'intermediate' question might be how to load biblatex with a class that has already loaded natibib. That might be trivial (it may just require removal of \ver@natbib.sty), but may not. – Joseph Wright Nov 27 '11 at 14:57
  • That's a good suggestion. Please note! This is hypothetical at the moment. I was wondering about the question long enough to begin wondering if it was sensible to wonder about the question. I'm not trying to sneak the question in here on meta. I'd be missing out on the rep points. ;) – qubyte Nov 27 '11 at 23:22

In my opinion, a question is only "too big" to be asked as tex.sx if it is a) in fact several questions at once or b) of the open-ended/subjective sort. A small, focused question that may lead to a bulky answer is perfectly fine, and your proposed question "How can REVTeX be made compatible with biblatex" seems to fit that description.

The worst that could happen is a question like Breaking equations with breqn with a large number of upvotes, but without any answer. ;-)

  • I find it very interesting that this answer has, at present, twice the upvotes of either of the other answers. I want to see how this changes for another day before marking any answer correct. Not that this actually has an impact on my decision. All of the answers have been useful. – qubyte Nov 28 '11 at 15:02

I would like to add a couple of things to this.

Firstly, the bigger the question then the more I want to see from the asker. The simplest is for the asker to show how much effort they have put in or are prepared to put in. For example, Ryan Reich asked a big question with How do I debug pgfkeys? in that since no-one already had an answer then an answer would involve a lot of hackery. But I know that Ryan loves PGFKeys so I know that he would put in a lot of effort into making it work. That gives me more motivation to help and I would feel that when answering that question I wouldn't have to give a full answer straight away, but start with some suggestions which Ryan could then take and then with some back-and-forth together it might lead to a solution. (Sadly, this is a hypothetical situation as I have too much else to do right now!)

Secondly, sometimes a question just happens to coincide with what I'm thinking about and I can see both that it would be an interesting answer to give and that others would benefit. Then I might write an answer that leads to a package. But you can't count on that happening so it's better to be in the first category.

(Note, this is in addition to the advice already given to break it down if possible. Sometimes, though, it just isn't possible to break it down any further.)

  • Good points. I'm glad I asked about this on meta first now. It's been very educational about the motivation of the people answer questions. – qubyte Nov 27 '11 at 23:24
  • People who answer questions. – qubyte Nov 28 '11 at 0:00

You must log in to answer this question.

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