koffice/krita/image

[Matthew Woehlke]

(repost with correct address.)
Um... please reply to kimageshop at kde.org. Sorry about that. (I'm trying 
to drop -commits from this, I don't think there is any reason to 
continue posting there?)

On 2009-11-05 03:39, Cyrille Berger wrote:
> On Wednesday 04 November 2009, Matthew Woehlke wrote:
>> On 2009-11-04 05:18, Cyrille Berger wrote:
>>> SVN commit 1044629 by berger:
>>>
>>> revert to 936510 (the reason is that after 944988, the algorithm return
>>> either a very small value or UINT64_MAX)
>>>
>>> It might be worth, at some point to backport to branch 2.1
>>>
>>> CCMAIL: mw_triad at users.sourceforge.net
>>
>> Um... 944988 should have been a non-functional change. Are you sure
>> something didn't go wrong since then?
> I tested 944988 first. And the change is rather intrusive.

Well, sure, it was a refactor after all. But it made the code cleaner
(and nuked a bunch of dead stuff). And it *shouldn't* have caused a
functional change.

(It did due to a cast that got lost, as noted on the list.)

>> Also it seems you reverted a number of compile fixes.
> Actually, I made sure that they were in my revert. But I did remove the ugly
> #ifdef and instead define the macro at the same place for everybody.

Okay. I'm starting with a clean revert, so I'll redo this separately
(assuming I can figure out what is the change - if you know offhand,
that would be appreciated). Probably I'll run that diff by you first.

Oh, and I am un-breaking the unit test :-). Right now it passes even
with the broken version.

Specifically, I am pretty sure you meant:
     quint64 number2 = randg.randomAt(...);
...and not:
     quint64 number2 = randg.doubleRandomAt(...);

(The *fixed* algorithm does pass the (fixed) test. This is good :-).)

On an unrelated note, svn is driving me nuts ;-). Does anyone have a
git-svn checkout I can clone?

-- 
Matthew