On 4/5/07, Charilaos Skiadas <skiadas@hanover.edu> wrote:
It's basically the viewer we are suggesting at the moment (once I
update the help file that is). It has the advantages of TeXniscope
along with some other ones of its own, and works on Intels much better.

Ah, okay. Unfortunately it doesn't work brilliantly with Watch, because its AppleEvents support really sucks, so there's no way to tell if the user has closed the document. Also, it insists on updating the PDF file whenever it's changed, and there's no way to externally signal it to update instead, so presumably it will sometimes detect the change while the PDF is half-generated, and break.

> I thought it might be useful to have the option of using a
> different previewer for watching than for 'Typeset & View', but
> perhaps no one would ever want to do that. At the very least, it
> could be an undocumented option, and we could say that the
> recommended way to change the viewer is to set TM_LATEX_VIEWER.

Yeah we could have it as an undocumented option.

I've removed it. If there turns out to be a need for it, it's trivial to put it back.

> Because of the non-standard name, I didn't realise this was an
> official configuration option. If it is, I agree that the name
> should be changed!

Well it's semi-official ;)  Supporting pstricks and the dvi->ps->pdf
route is relatively new and hasn't settled in much. So TEX_PSTRICKS
was really at that point just an internal variable defined within the
LaTeX&View command, not supposed to be settable by the user really.

That's what I'd imagined.

I wonder if it is necessary at all. Is that really different than
setting the TS-program to "latex" instead of "pdflatex"?

No, not at all. But it's nice to have a global option, so you don't have to edit every single source file when you want the same for all them. But it should be possible to override the global default in the file.

I guess this is the question: Should the user have to specify
anything else other than whether they want pdflatex or latex->dvi->ps-
>pdf ? Ideally one variable should do the trick, and even that
should only be needed rarely.

I heartily agree with the sentiment that it should be as simple as possible. (And maybe one day someone will write a nice graphical interface for the configuration.)

But it's not *quite* this simple. People do use xelatex, for example. Perhaps we should just use the same names as TeXShop does (so people don't have to learn two different ways of specifying the same thing), and have a variable TS_LATEX_PROGRAM that can be (at least in principle – I guess not all the options will be supported immediately) set to the same things that you can put after TS-program =.

So, for example, TX_LATEX_PROGRAM=latex would force the DVI route unless it's over-ridden by an explicit %!TEX line in the document, etc.

What do you think of this idea?

Hm, that's a bug in the way we do these things currently I would say,
the %!TEX directive should probably be taking precedence. Basically
the whole setting sniffing method in the Typeset&View command needs a
good cleanup. It has been patched so many times that it is somewhat
inefficient/buggy atm.

I don't mind fixing up this script to support the mechanism I suggested above, if you agree it's a good idea.

Yes I meant to make the binary available somewhere, and then link to
it from the documentation. I was thinking of someone's webpage. I can
put it on my blog somewhere, or you can host in on your homepage.

I don't really have a proper homepage at the moment. Well, I have my university one, but that will go away fairly soon, and I have a small amount of webspace from my ISP (where I put the 2.0 bundle), but I don't know how long I'll keep the same ISP.

If you have room for it on your blog, perhaps that would be the best place.

> GPL.

Then I am not sure we can actually include it in a bundle that ships
with TM, or include it at all for that matter. Can we?

I hope so! CocoaDialog is GPL, and that's included.

I'm pretty sure it's okay. The GPL specifically excludes "mere aggregation" in its viral clause.

Robin