Alexandre Julliard wrote:
Ivan Gyurdiev ivg231@gmail.com writes:
Changelog:
- Abandon central memory allocation, and the tracking that involves.
Write a setup() and teardown() handler for each test, and use heap allocation.
- Make the main test function configure test arguments.
Replace get/set arguments with a single test_arg.
- Remove some const qualifiers on test data.
Const qualifiers should go on function parameters instead.
When you find yourself writing a Changelog like this, it's a clear sign that it should be 3 separate patches.
Well, sort of...
I want to add a setup() and teardown() handler.
To do that, I need to give them an argument - it makes sense to make that argument shared with the other functions for simplicity (kind of like a context to the handlers). Now I find that a bug in const handling is causing compiler warnings all over the place (in process of writing a setup/teardown handler), so I'm forced to fix that too.
Essentially the first change drives the other two. Sure, I suppose I could revert the changelog and send #2 and #3 first, then make the change I'm interested in. However, I'm not sure why that would be better - it would accomplish the exact same thing in twice the amount of time. The tests pass before, and they will pass after as well.
Ultimately this patch isn't what I'm interested in anyway - I'm looking at new tests (already written), which expose quite a few new bugs. I just need to figure out a refcounting issue before my test will terminate properly. Your approach is more appropriate from an observer perspective, but forces me to work on breaking down patches, instead of writing new functionality.
==========
Were those patches dropped for a similar reason ?
[WINED3D] Pbuffer/ActiveRender cleanup #2 [WINED3D] Clean up some shader issues