[reportlab-users] Re: The upcoming Reportlab 1.18 release and patches.

[Robin Becker]

In article <ur850p7qs.fsf@boost-consulting.com>, David Abrahams
<dave@boost-consulting.com> writes
>Robin Becker <robin@reportlab.com> writes:
>
>> In article <uof04sivh.fsf@boost-consulting.com>, David Abrahams
>> <dave@boost-consulting.com> writes
>>>
>>>Good, that's the problem.  I've thrown valid XML at reportlab and had
>>>it choke on its own 4-element return from parsexml0 when it gets
>>>assigned in reportlab.lib.rparsexml.parsexmlSimple.
>>>
>> .... yes the problem was me thinking you said the two element return was
>> at fault so leading me astray. This is not my code and I have never
>> looked at it seriously unfortunately I don't use it either.
>
>AFAICT it's pretty much impossible to avoid if you use platypus at
>all.  Para decides to use parsexmlSimple based on some very trivial
>tests on the XML it's passed.  The easy fix is to disable the use of
>parsexmlSimple as I've done in my copy of the code.
>
para is not the default paragraph implementation so I can avoid using it
completely. There are difficulties with both implementations, until
yesterday you couldn't get '&amp;amp;' to produce '&amp;' in para, the
older paragraph can't do underlining etc.

In fact para itself is not required to use the parsexml0 stuff, the
latest pyRXP could probably handle it directly with the right flag
settings, although given the speed of Aaron's parser it probably
wouldn't provide much speedup.
-- 
Robin Becker