On Friday, 9 February 2024 11:18:34 CST DodoGTA GT wrote:
Hello,
I have made an updated version of the reparse points patchset fix that should reduce the patch noise quite a bit (I think the original version is better in terms of code duplication though)
Thanks, this makes it clearer what you're trying to fix.
I've rewritten the fix for patch 5 to properly handle delayed closes. It's hard to read the diff but I suspect your original patch did that, and was more correct in that respect.
Unfortunately reading diffs of diffs is very hard (even with practice), and reviewing them is an order of magnitude harder. In the future, if you have a change for wine-staging to submit, I'd ask that it either be in the form of a first-level diff against the Wine tree, which we'll apply as a separate patch or merge into the correct patches as appropriate, or simply a description of the change to be made. E.g. in this case "unlink() was incorrectly replaced with unlink_closed_fd(), when instead the diff to inode_destroy() should have been moved inside unlink_closed_fd()". I know the latter seems well against the usual practice of free software collaboration, but it really is easier for us to deal with.
I've also removed the extraneous parts from your other patch, and applied the remaining parts to the relevant patches.
Thanks for the patches.
--Zeb