Re: [Boost-docs] Re: can't build QuickBook documentation

[James Fowler]

João Abecasis wrote: Ok. I think I've got this fixed now. I was missing a path.native on the quickbook command. >From the last changes I've made to the toolset all that should be *required* now is:     import quickbook ; in the Jamfile using quickbook. Maybe that one-line could be folded into the boostbook toolset and ours becomes just an implementation detail. Thoughts? (and further bug reports...) yeah I got some thoughts...  how about we fold the boostbook toolset into the quickbook stuff and make boostbook the implementation detail... maybe make it disappear altogether for a while...  sorry, I'm a bit grouchy ;)  I've spent most of the last two days fighting with the doxygen2boostbook.xsl, and I'm almost frustrated enough to try rewriting the whole thing in python. Maybe it's just me, but when it takes hours of trying to debug a few pages of XSLT which could have been about 30 lines and 15 minutes of perl or python, something just ain't right.  Of course, maybe basic XSLT skills will be like vi - better than nothing, if only because of ubiquity.  Anyhow, it's nearly been bludgeoned into a usable form for doxygen 1.4.1 (namespaces, nested classes, gets parameter descriptions on functions & macros) but it just shouldn't be this hard to do such simple stuff.  If any of you are brave (yeah, thats it...) enough to try setting up doxygen 1.4.1 with boostbook,  let me know and I'll send you the updated doxygen2boostbook.xsl. Another thing:  I modified my quickbook.jam to handle [include ...] dependency checking.  It mostly works, which means that it usually catches changes properly and triggers a rebuild.  Builds in general (not just for qbks) with bjam --v2 seem to occasionally miss changed dependencies on my system. # quickbook-scanner finds [include ...] dependencies so BoostBuild knows # when a modified included qbk means we need to rerun QuickBook # class quickbook-scanner : common-scanner {     rule pattern ( )     {        return "\\[include[ ]*([^\]]+)\]" ;     } } scanner.register quickbook-scanner ; type.set-scanner QUICKBOOK : quickbook-scanner ; Not sure why it wants the _expression_ to start with "\\[inc..." instead of "\[inc...", but that's the way bjam likes it (according to "bjam --v2 -d+3" which shows the dependencies - useful for sanity checks on scanners). - james -- __________________________________________________________ James Fowler, Open Sea Consulting http://www.OpenSeaConsulting.com, Marietta, Georgia, USA Do C++ Right. http://www.OpenCpp.org, opening soon!