Re: [boost] New Acknowldegement for RenderX

From: John Maddock (john_at_[hidden])
Date: 2007-08-19 04:35:37

Rene Rivera wrote:
> Taking this to the docs list...
>
> John Maddock wrote:
>> Doug: is there server space we can put this on?
>
> I brought the possible need for a documentation building server with
> the moderators last week. Which includes you John ;-)

Um, so you did, now where did I leave that brain of mine....

> I mentioned
> setting up a buildbot for this. In particular it would be possible to
> have a doc building slave which others can ask to build docs with
> (from the buildbot web interface). If the doc builder is the OSL
> machine the docs can be made available through beta.boost.org.
>
>> The tricky part I suspect
>> is getting links to images to function correctly, we may need some
>> adjustments to BBv2 before we can make full use of this?
>
> Which images?

Here's the outstanding issue: suppose I have a quickbook file that 
references some png's (or whatever):

[$../graphs/graph1.png]

The path ../ is set up so that it's relative to where the html docs are 
generated (in a html/ subdirectory), but FO files and PDF's currently end up 
in a long mangled path under bin.v2/.  So... neither admonishment graphics, 
nor referenced image files are found, unless you manually move them to where 
the FO processor is looking for them.  Not fun :-(

So the question is how do we get the server to "work".  If we instruct it to 
build from a Jamfile in SVN, then IMO we need to fix up BBv2 so that .fo and 
.pdf files get placed in a sub-directory (called pdf?) just like HTML does, 
and then hopefully external image files will get automatically found.  In 
fact I believe this is the "right thing to do" anyway, as it eases local 
building.

Alternatively, the easiest option for the server setup might be to only 
accepts self-contained .fo files with all the images embedded inside it. 
Unfortunately I don't believe that XSLT processing can automatically embed 
externally referenced image files, so that would need a post-processing step 
or something.

Thoughts?

John.