Re: [AD] Bug with transparent ellipses

[Jon Rafkind <workmin@xxxxxxxxxx>]

Elias Pschernig wrote:
> On Mon, 2006-09-04 at 11:59 +0200, Evert Glebbeek wrote:
>   
>> On Monday 04 September 2006 11:49, Elias Pschernig wrote:
>>     
>> I don't think we should be doing (relatively) fancy things like this so short 
>> before a release. Just stick ASSERTS in the appropriate places for now.
>> Also remember: 4.2.1 is a bugfix release, not a release to cure the ills of 
>> the current API. Changing the code so that the colour mapping tables are not 
>> NULL to begin with can be argued to be an API change, which might break 
>> existing code. I don't think we want to go there.
>> I'm ok with discussing this for a possible 4.2.2, and it should definately be 
>> done better for 4.3 (but that's the idea anyway, right?), but not for 4.2.1
>>
>>     
>
> Yes, that's just what I thought earlier.. the problem is, in #allegro it
> got clear that patch to place ASSERTs would be slightly harder to do and
> look messier than this one.
>
> And so, as we now have a very simple and almost (remove the rgb_map
> stuff) working patch, lazyness would win for me :)
>
>   

 It did seem to me that it would be riskier to fix the macros since its
not entirely obvious that would fix every case, and its sort of tedious
to fix them all. Using default values guarantees segfaults wont occur
and although there are a few strange cases like the one with rgb_map
they are easier to spot.

Also maybe slightly off-topic but I dont find ASSERTs that valuable.
When my program segfaults the first thing I do is not re-run it with a
debugged version of Allegro, rather I pour through my own code to see
what I messed up( since I trust Allegro that much :) ). Perhaps I should
use the debugged version instead but I imagine many other people don't,
especially people with less experience.