> to disable xmltex after all, just figure out how to avoid the problems while 
> building, and comment that both routes do not yet produce nice output.

if only life was that simple..

first, the passivetex problem with computed lengths seems well known, and
has already been (partially) addressed in docbook, eg:

http://sourceforge.net/project/shownotes.php?release_id=287017

2004-11-20  Michael Smith <[EMAIL PROTECTED]>
  | 
  |  * lists.xsl: bug #1007414 & patch #971730 - passivetex outputs
  |    "* 0.60 + 1em" garbage in lists (reported by Alex Roitman
  |    and James William Pye) ::PROBLEM Literal stuff "* 0.60 +
  |    1em" showing up in listitem PDF output generated through
  |    passivetex ::CAUSE Logic that outputs non-constant size
  |    expressions like "4em * 0.60 + 1em" was added to stylesheets
  |    to improve look of list formatting. But passivetex can only
  |    handle constant sizes; so, instead of doing the math, it
  |    spits out the literal expressions in output. ::FIX Added
  |    workaround - some conditional logic to fall back to constant
  |    values if passivetex.extensions param is non-zero.
  |    ::AFFECTED AREAS Should only affect FO output when
  |    passivetex.extensions is non-zero

but unfortunately, adding '--stringparam passivetex.extensions 1' 
to the XSLTPROC call does not entirely solve the problem, because 
one of the other offenders (/usr/share/docbook-xsl/fo/param.xsl, 
responsible for the '<fontsize> * 1.2' kind of things, tries to do
the calculations in xsl. in principle, this is better than the 
questionable workaround above (which simply omits the 
calculations, and does so depending on a flag intended for 
other purposes..), but it still produces some computations as 
strings. so xmltex will still hang at those points.

i don't want to start patching docbook, but the perl-hack isn't 
really a better "fix", either, and the changes are fairly simple (just
replace string concatenation by element concatenation in two
more places, to enable evaluation at xsl-level (diff attached,
send whereever appropriate, please), as suggested here:

http://lists.oasis-open.org/archives/docbook-apps/200412/msg00048.html

while searching for this, I also found that these extensions are 
supposed to enable pdf toc bookmarks, so it might be good to 
switch them on anyway. it does seem to work nicely with fop 
and adding '--stringparam fop.extensions 1' for XSLTPROC.

(switching on extensions for one tool seems to confuse the other,
even if it shouldn't, so this should be configuration-specific; if 
you see "fixes" like the one at the top, that is perhaps not surprising)

at which point I guess I better get back to my other problems,
and leave this one to one of you;-)

summary so far (hope I've not missed anything?):

- fix table column declarations in users_guide/flags.xml

- if using xmltex/passivetex, 
    - check and if necessary patch /usr/share/docbook-xsl/fo/param.xsl
    - enable passivetex.extensions for xsltproc (to get the workaround)
    - avoid '_' in xmltex input file name (eg., use users-guide.fo)

- if using fop 
    - look for fop and fop.sh in configure
    - enable fop.extensions for xsltproc (to get nice pdf toc bookmarks;
        also seems to remove many of the error messages?)

- do not enable extensions for one tool while using another

cheers,
claus

Attachment: param.xsl.diff
Description: Binary data

_______________________________________________
Cvs-ghc mailing list
[EMAIL PROTECTED]
http://www.haskell.org/mailman/listinfo/cvs-ghc

Reply via email to