* Mads Toftum wrote:
As it is I'd much prefer to have xml only for all generated pages in cvs and only create html when rolling the tarball.
That's not so easy like it sounds.
It assumes, that the RM has to be properly installed java runtime and the neccessary classes (xalan, xerces, ant etc.).
The other point is, that he should control all generated files (currently about 200). For example, I have the problem, that I'm not able to build the japanese html files properly (have some weird problems with special characters). And controlling is not possible anyway since I simply cannot read japanese ;-)
Last but not least, the docs have to go some way to the online site (httpd.apache.org/docs-x.y). Either one would build them locally and put them up or build them there on the server (also a properly installed build system assumed, see above). Currently it's a simple cvs checkout.
I would also like to see only the xml sources in cvs (I hate the nasty HTML update mails), but for the reasons Andre pointed out, it would be to difficult and complex.
for the online site I think we could use nagoya (has a sufficient JVM AFAIK) to build the docs, and then copy them over to daedalus. but that's only one point.
However, we could use separate directories for xml and html, like the httpd-site does. I would support that.
this sounds much better to me. Perhaps we should brainstorm some possible solutions for this. I would like to see some layout in the following form:
| | - build | - docs-html | - developer + - other html-only files | - docs-xhtml | - howto | - mod + - *.html | - docs-xml | - howto | - mod + - *.xml | - docs-xxx (pdf, rtf, ...?) | - howto | - mod + - *.pdf | - search | - share | - images + - style or (my favourite, if it wouldn't be too hard): | - css | - dtd | - img + - xsl
Just an idea :-) Don't have the time to implement that right now, but it would be much cleaner to work with...Any thoughts?
cheers, Erik
