On 10.06.2015 10:38, Alexandre Julliard wrote:
What should happen is that Qian would submit a patch (to wine-patches, where everybody can see it) then if you see something wrong with it, instead of going ahead and making changes yourself, you tell Qian about it (and CC wine-devel so that we are all aware of what's going on), and let him address the issues and resubmit the patch. Repeat until the patch is good enough.
Hello Alexandre,
It is a bit unrealistic that from now on everyone is going to submit their raw draft patches to wine-patches, although they need additional work.
Moreover, as mentioned in the first mail, we were working together on fixing various msys2 bugs. If Qian would have preferred to tackle this problem himself, I would have given him some additional time of course, before I start with my own attempt. @Qian: Please correct me if I'm wrong, and you feel like I'm stealing your work.
The process of having people fix their own patch is important. Not only it lets them learn from their mistakes, it also enables us to see where they are struggling, how they improve, how much they care, and builds the trust necessary to get patches approved. The process is at least as important as the final result.
That's my issue with wine-staging: it's trying to shortcut the process on the idea that if the final patch is OK, it doesn't matter how you got there. But to me, it does very much matter.
I have to admit that the whole idea sounds nice, but based on my experience it doesn't really work very well. Such a method only works when every patch gets detailed and fair feedback, which is not the case at the moment. A lot of patches are not reviewed at all. I can perfectly understand that a lot of longtime wine developers don't want to mess around with patches of a very low quality, but what exactly should people learn from being ignored?
For others the feedback is sometimes a bit too strict if you ask me. Trying to submit an unfinished patch on the mailing list, will not result in an answer "let us work together to get this upstream", but often a kinda harsh and almost unfriendly answer, which leaves the impression that noone really cares about it. Also, in situations where it is not reasonable to ask for another iteration, or where the author has given up, it should be possible to integrate the last available version, to allow others to continue to work on it, which is also not the case right now.
Regarding the concern that Wine Staging makes it impossible to follow the process how the patches were created - do we know how many iterations developers do in their private repositories, before they finally submit it? Do we know how often people ask for feedback privately, before sending their patch? Obviously not. If there is a fear that upstream developers miss anything important, we should think about integrating Wine Staging a bit better into the upstream concept of submitting patches, instead of blaming it for everything. Wine Staging is not the problem, its an attempt to solve it, and only future will be able to tell if its the right approach.
Best regards, Sebastian