[pmwiki-users] Bibliographies

[christian.ridderstrom at gmail.com]

On Mon, 11 Sep 2006, John Rankin wrote:

> I bow to your knowledge in this area. When we looked at whether to turn 
> wiki markup into LaTeX directly or use an intermediate XML format, we
> observed the following:
> - it's not the markup that gets you, but the tables (sometimes these
>   have to go into separate LaTeX files), the HTML entities, the images
>   (these all have to be converted), the print metadata (what font to use,
>   whether the document is single or double sided, and so on), the
>   equations, and such like

Yeah... I can imagine that all these extras are a real bitch. One of the 
strengths of LyX is that it has tools/scripts for converting images into 
suitable formats. The whole package of files are then moved into a 
separate, temporary directory, where LaTeX can finally parse the files. So 
I understand that your wikipublisher is doing a *lot* of stuff more than 
just generating a .tex-file. I assume you have to do all of the above...

> - wikistyles are really, really hard to handle -- mapping styles
>   to their LaTeX equivalents is a challenge that we only partially
>   overcame (example: coloured text that contains a url can in some
>   cases cause LaTeX to die)

Is there some mechanism for definining the LaTeX equivalent of a 
wikistyle? (Otherwise I probably wouldn't even try to support more than a 
small subset of them). Hmm... maybe you could add a method that let's 
people define the LaTeX equivalent of a wikistyle?

> - using an xml format means that you can detect (and discard) any 
>   spurius html tags that you didn't manage to catch (we use a namespace 
>   to qualify all the tags we insert, then remove any tags without a 
>   namespace, then remove the namespace) -- this is especially important 
>   when using third party cookbook recipes, which may insert hard-coded 
>   html tags

Clever!

> In the end we decided that pmwiki was designed to output xml and we were 
> on safer ground by choosing to output xml. But I'm aware that this was a 
> judgement made with very imperfect information and there may be a better 
> way.

Oh... just to avoid misunderstandings, I never meant to imply that 
wikipublisher shouldn't output XML. To me that sounds like a very good 
choice. I only meant that if we don't care about images and other 
complicating stuff, it's not that much difference in outputting the lyx 
format immediately.

> I thought of yet another option: take the wikibook xml that we generate 
> and use xsl to transform it to Lyx. I suspect that it would be 
> relatively straightforward (if time-consuming) to modify the .xsl that 
> converts the xml to LaTeX.

That's probably not so difficult as you say... but on the other hand it 
seems a bit silly as LyX *will* move to an XML fileformat. And even if LyX 
will be able to convert old .lyx-files (being backwards compatible), it 
seems a bit silly...

/Christian

-- 
Christian Ridderström, +46-8-768 39 44               http://www.md.kth.se/~chr