Re[2]: [lmi] notes on using xmlroff

[Vadim Zeitlin]

On Tue, 12 Jan 2010 00:00:46 +0100 Vaclav Slavik <address@hidden> wrote:

VS> > IOW, if we began afresh today with the goal of creating formatted,
VS> > paginated pdf output from text and numbers created in C++, would we
VS> > pick a different path than C++ --> xml --> xslt --> fo --> pdf?
VS> 
VS> I may be missing something, but why does LMI use XML step in the first
VS> place? Does it serve any purpose other than being XSL:FO input? 

 This was exactly the question I wanted to ask too. I seem to remember
there was some reason to do it like this but I just can't find it any more
now. I do remember the reasons that were given for using XSL:FO (in brief,
the precision of the output) but I suspect we could do just as well by
directly generating PDF.

VS> Personally, I'd probably go with direct use of some reporting library
VS> with PDF output, such as ReportLab. ReportLab is for Python; I know
VS> there are some C++ and plenty of Java ones, but have no idea about their
VS> quality.

 There is even a wx solution (which I didn't use but heard a lot of good
about) at http://wxcode.sourceforge.net/components/wxpdfdoc/ but maybe it's
not appropriate if we want to generate PDF from non-wx code too. Anyhow, I
certainly am under impression that it is possible to generate PDF directly
without too many problems, especially as long as it's relatively simple,
please let me know if you'd be interested to know more about wxPdfDocument
or any other C/C++ PDF generation libraries.

 And if we had to go via an intermediate representation, I'd choose
PostScript as it's easy to generate and debug and can be converted to PDF
easily too.

 Regards,
VZ