[reportlab-users] More bugs with Paragraph, with underlined links

[Dinu Gherman]

Andy Robinson:


> As a pragmatic atheist, I suggest coding instead of praying ;-)

>

> Paragraphs are unclean internally, but very well tested in a wide

> range of conditions, and changes here are highly likely to disrupt

> almost every application out there.  So we don't want to mess with the

> existing implementation just now.


This is too vague a statement to be of any use for me, sorry to say  
that!
I haven't (of course) seen tests in the test suite covering these  
errors.
The last one I reported is *very* easy to detect. The one before is  
hard,
I agree.

I fail completely to understand, why fixing a bug "disrupts almost every
app out there". It should fix them too, unless you actively used work-
arounds in every other app - which would be a waste of time, that should
better have gone into bug fixing.


> But if you or someone else can produce something new which has a clean

> algorithm to pack and split inline objects, we'll certainly look at it

> and see if we can help.  I suspect that the right way is to make a

> clean linear packer and then rewrite it in C.


Concerning clean paragraphs, I suggest you write down a couple of use
cases and I'll give it a try. It might also be useful as a start to try
to describe in plain English, what the current Paragraph class tries to
to, if you want people to contribute instead of being scared off by the
source.

Something like this helped me a lot with my own code for converting  
Open-
Office ODT to ReportLab Platypus, where my first attempt was a total  
mess,
but then improved greatly, so I could sleep well again - if only the tag
bugs would disappear. How about RL fixes its paragraphs and I contribute
my OpenOffice converter? That sounds like a good deal to me... ;-)

Regards,

Dinu

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://two.pairlist.net/pipermail/reportlab-users/attachments/20071108/7e59b4eb/attachment.html>