[reportlab-users] ReportLab 2.6 Bugs

[Stephan Richter]

On Friday, December 21, 2012 02:23:49 PM Robin Becker wrote:

> We do have some tests for various Pie3d charts, but those aren't in the

> main  tests. Most of those are to do with testing for bitmap images leaking

> fills at small angles etc etc (ie for failures in the bitmap rasterizer). I

> made the obvious change to Pie3d to fix this issue and those are passing

> still so I will commit all of your recommendations today.


Great! BTW, I just ran coverage against reportlab from my z3c.rml code and the 
coverage is at a dismal 39%. That said, there are a lot of uncovered lines 
because a lot of modules have manual tests and samples at the end of the file, 
which are not hidden in "test" or "tests" modules. So maybe that can be 
addressed somehow over time.

Also, now that we have identified all module-global variables that prohibited 
repeated renderer use, we are in the position to make the library thread-safe. 
There are 2 solutions:

1. The simple one: Make all those variables threadlocal variables.

2. The involved one: Create one global Configuration object that holds all the 
variables and that is passed around for one rendering. (That config object 
could also be stored in a threadlocal to make the task easier.) 

What do you guys think?

Regards,
Stephan
-- 
Entrepreneur and Software Geek
Google me. "Zope Stephan Richter"