do i add a special check in buildtools to not generate duplicate implicit conversion operators?
do i add a special check in buildtools to not generate duplicate implicit conversion operators?
Lits 3am, ill finish this tomorrow if you dont want to do it/cant, the rest should be relatively simple "copy structs from buildtools output, fix small buildtools bug"
Pyeah i won't have the time until friday so all yours
Pbut that's good progress!
Pi'm not objected to adding more headers but if we go down a header spiral it gets crazy
PpDispParams and pVarResult could in theory be void*s
Pbut i feel like given those are POD structs we should just generate them
Lalright while im fully awake and have the energy to, before i continue further im going to go through and clean up this gross git history
Lcommit chains like this dont really need to exist, and if they do, they should be named correctly
Lthose 3 commits can be merged
Lthis commit can be dropped (its already upstream)
Lthis can be dropped
Lalright
Lthat should be a lot better
Loh f5265d can be dropped here as you fixed that already
Lokay that should be good now
Lthis should be the full new commit history
Lmerged as many related commits as possible, and dropped commits that were duplicated work/already fixed upstream
Lalright i have finally fully fixed it and done a cursorary glance to it with
git log --patch and aside from it crashing a lot from the large commits it seems actually managable now hopefullyLive never had a git history cry this bad
Li think its the squash
Lcus i cherrypicked/rebased a few times and never had problems like these
Lbut since the commits that i cherrypicked/rebased were lost in the squash
Lgit got confused
Lnow i know to be actually careful :^)
LMUCH better
Plol this is why you never force push
Lthe problem was more the sqash than the force push, this still would have happened if i merged instead of rebased
Lthe problem is the sqash confusing git
Lluckily its fixable
with a rebase
with a rebase
Kneeds more
git push -f
Pthe trick that worked for me last time is dropping the commits from the rebase, pushing, resetting to origin/main, merge the origin/branch into branch, then push
Li think my solution is just to not do cross-PR shenanigans next time (it also gets very overwhelming very fast)
Lalright this shoudl should be FINE
Lhopefully CI doesnt cry about it
Lit builds locally
PWill do a review this evening and update the blog post
Poh wait @baebey one thing we do have a precedent of is omitting the tag prefix when renaming these structs as this is what the typedef does in C land
Pthat’s the only thing I remember seeing in a super quick skim at work
Phappy to update that as I go if you’d like
P(That offer is only good for like 10 minutes though)
Pnevermind, i don't think you committed the generator.json - yields a completely different output on my machine
Pafter that, and the rename of the tag structs to not have tag in them (except for TagDecimal because it'll conflict with the BCL) it is good for merge
Pand 2.17 is good for release
KThis demo looks so clean, I love it!
Kalso this big one
KI want to turn it into a demo somehow, it looks so cool
Kmaybe some kind of liquid animation in the background, or cloth 
Lwait i thought the textured quad was supposed to have a green background