Uh oh!
There was an error while loading. Please reload this page.
deps: update V8 to 7.4 - #26685
Conversation
ryzokuken
commented
Mar 15, 2019
targos
commented
Mar 15, 2019
/cc @nodejs/v8-update |
addaleax
commented
Mar 15, 2019
@ryzokuken@targos If you want, feel free to include 3ba72b5...51a9301 (v8-7.4 branch on my fork) – all of them are clean cherry-picks of upstream commits, and bring us to API + ABI backwards compatibility with V8 master. I can also push them directly to this branch, if you prefer. (Adding v8/v8@de7ab39 to that would actually bring us to 1:1 identical APIs, but it’s purely additive and would be just part of a V8 7.5 semver-minor bump on our side.) CI for those commits: https://ci.nodejs.org/job/node-test-commit/26781/ |
refack
commented
Mar 16, 2019
@ryzokuken would you consider adding |
ryzokuken
commented
Mar 16, 2019
@refack I'd be more than happy to do so. That said, I personally believe it's more suited to the case where commits are made to a repository in org X by a person representing org Y, thus on behalf of the entire organization. For instance, I think upstreaming patches to V8 can be done using this. Also, I don't think this applies for me anyway:
|
refack
commented
Mar 16, 2019
Yeah, it's worth thinking about and discussing some more. As I see it, it's useful as a way to disclaim changes that are not original work by the commiter... Actually the designation should probably be |
ryzokuken
commented
Mar 16, 2019
@refack this sounds much closer to the real situation, and I wholeheartedly agree on discussing it further. For now, I'll add |
ryzokuken
commented
Mar 16, 2019
BTW, I just realized that the following don't hold up for me:
Because of these, I don't think I can make commits on "behalf of" V8. I will pick this up with the V8 team. |
refack
commented
Mar 16, 2019
Fixes: #26694 |
ryzokuken
commented
Mar 16, 2019
targos
commented
Mar 16, 2019
segfault on LinuxONE. Maybe because of f1d3b41 ? |
targos
commented
Mar 16, 2019
Yeah, we probably need to cherry-pick other commits. There are some with a message like " Prepare for changing kTaggedSize" |
refack
commented
Mar 16, 2019
FYI: I'm strongly considering deprecating the use of pure numbers as values for GYP3 (I want it to be pure strings)... defb(opt):
return'true'ifoptelse''this way in the conditions: [
["opt", {}, {}],
["not opt2", {}, {}],
],but I need to think about this some more. |
refack
commented
Mar 16, 2019
BTW: what the tl;dr about the timeline for landing this? |
targos
commented
Mar 16, 2019
ASAP to have it in v12 release candidates |
targos
commented
Mar 16, 2019
I cherry-picked all remaining commits related to v8:7703 (and v8/v8@4e6a1a7 to avoid a conflict) Hopefully we're good: https://ci.nodejs.org/job/node-test-pull-request/21595/ |
targos
commented
Mar 16, 2019
Weird, both Windows builders failed git commands. macOS and AIX were already failing before the cherry-picks. |
targos
commented
Mar 16, 2019
BTW, this version of V8 breaks our V8-CI because: |
V8-CI issue is nodejs/node-v8#100 |
ETA: April 23rd, 2019
Intended to supersede #25852
/cc @targos 🎉