-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Am 2014-11-27 14:14, schrieb Joachim Priesner:
Am Donnerstag, 27. November 2014 schrieb Stefan Dösinger:
This is written from scratch, but inspired by the patch from Joachim Priesner. The included test tests the absolute minimum necessary to show that the abs() needs to be placed in the fixed function vertex pipeline replacement shader. It ignores vertex and pixel shader behavior.
Does this mean I can finally stop worrying about my version of the bug fix and concentrate on getting my other two pending patch sets in?
Yes. Version 3 of my patches made it in, so your game should work now. The part from your test that tested the vertex shader behavior is still on my todo list, but since we don't know any application that has a problem with the current behavior it is fairly low priority.
I must admit that the way we handled this was not our finest hour. It is quite clear that we have a big problem with handling new or sporadic contributors.
ws2_32: WS_get_local_ips: Fix ordering of IP addresses by metric if two addresses have the same metric (try 2)
which seems pretty straightforward and fixes a real bug, and also on
I agree that it is straightforward. I only see a minor issue: I think you can drop the if (!left || !right) check - the qsort man page suggests this is not necessary.
I guess writing tests for this is impossible because the input comes directly from the host system's network setup which can only be modified with root permissions.
The other problem ws2_32 has is that there's apparently no one except Alexandre who claims to know it well.
[PATCH 3/3] comctl32: Implement the TaskDialog function (try 2) [PATCH 2/3] comctl32: Basic implementation of TaskDialogIndirect (try 2) [PATCH 1/3] comctl32: Move TaskDialogIndirect function to separate file (try 2)
I have never touched comctl32, so I'm not really qualified here. The overall feeling is that the code looks like is is copypasted together from various places. (The style inconsistencies trigger that suspicion). Combined with the lack of tests it raises the question of how solid this implementation is in the long run.
Also the defines like TD_WARNING_ICON look like they're copypasted from Microsoft's header.