v5.8.0 proposal #5559
v5.8.0 proposal #5559
Conversation
|
@rvagg @jasnell @trevnorris @Fishrock123 It was discovered in #5555 that d77c3bf has introduced a pretty nasty regression... specifically that calling to My notes on this can be found --> https://gist.github.com/TheAlphaNerd/6615a27684deb682dfe7 @chrisdickinson was speculating that a fix will likely involve the code found --> #4507 If there is a quick fix that we can get out then we should definitely close this proposal... but if we can't get it out ASAP I think we should move forward with the reversion on v5.x and this release. The regression came in the same release as the fix to the latest openssl bugs, and as such we currently do not have a release on the v5 line that is secure and working. |
mscdex
added
meta
|
CI: https://ci.nodejs.org/job/node-test-pull-request/1843/ edit: Citgm failures are either infra or expected |
|
I don't currently know of a quick fix for this. @indutny have an idea? |
|
Can we please do this properly and land it upstream first? |
|
@jbergstroem the idea here was to hotfix this in the v5.x branch so that we can have a working release while not reverting the change on master giving it time to be fixed properly |
|
(if this is actually an issue upstream, sorry - changes looks related to how we use it) edit: sorry, not relevant upstream indeed. |
|
@trevnorris neither do I right now, will take a deeper look. |
|
+1 to reverting from v5.x and getting a quick release out. |
|
#5585 is also a new |
|
@rvagg mentioned to me that him and @jbergstroem were reworking the release CI but didn't get it done yesterday, so it might not work for now. Let's just do a regular release tomorrow; I'm willing to pick it up and do it if you'd like. |
MylesBorins
force-pushed
the
v5.7.2-proposal
branch
3 times, most recently
from
da768c7
to
5de263e
Compare
|
@thealphanerd fix for http regression: #5591 |
|
@Fishrock123 the new ci should be done by today; already doing test builds. |
|
Here's a list of proposal commits (only waiting for #5591):
|
|
@Fishrock123 would you like to take over this release? Initially this was going for a quick knee jerk release last week, but since we've delayed until tuesday it would make sense to me for you to take over |
|
looks like we might be missing some [079973b] - deps: add V8 DEP trace_event (Ali Ijaz Sheikh) #4722 /cc @ofrobots the rest look good to me |
|
@Fishrock123 we likely should include the latest npm fix as well
edit: I just landed this |
|
Yeah I'll take it over for tomorrow (pending the http bugfix), no worries. |
|
@ofrobots cool, thanks. If you tag it with the |
|
Does the V8 bump to 4.9 not mean we bump minor? |
|
there is no v8 bump, we're committed to keeping v8 stable on v5, that should change in v6 but I think we need to do some documentation work on this cause it mostly exists in people's heads |
Fishrock123
force-pushed
the
v5.7.2-proposal
branch
from
5de263e
to
07e6fee
Compare
|
osx should be happier now if you want to run CI again |
|
Sure, why not. Just finished the changelog, will also be making release builds soon. |
Notable changes: * child_process: “send()” now accepts an options parameter (cjihrig) #5283 - Currently the only option is “keepOpen”, which keeps the underlying socket open after the message is sent. * constants: “ENGINE_METHOD_RSA” is now correctly exposed (Sam Roberts) #5463 * Fixed two regressions which originated in v5.7.0: - http: Errors inside of http client callbacks now propagate correctly (Trevor Norris) #5591 - path: Fixed normalization of absolute paths (Evan Lucas) #5589 * repl: “start()” no longer requires an options parameter (cjihrig) #5388 * util: Improved “format()” performance 50-300% (Evan Lucas) #5360 PR-URL: #5559
PR-URL: #5559
Fishrock123
force-pushed
the
v5.7.2-proposal
branch
from
07e6fee
to
8165570
Compare
|
Release ci attempt: https://ci-release.nodejs.org/job/iojs+release/461/ Edit: now at https://ci-release.nodejs.org/job/iojs+release/463/ (maybe?) |
|
@jbergstroem check out @Fishrock123's release build attempt @ https://ci-release.nodejs.org/job/iojs+release/461/ - it's assigned to @thealphanerd doing v5 which actually occured @ https://ci-release.nodejs.org/job/iojs+release/462/, got any clue how this might get mixed up? |
|
@rvagg 461 and 462 are v4.4.0 builds. it looks like @Fishrock123's build is 463 |
|
@thealphanerd right, but @Fishrock123 swears that he started 461 4 hours ago but now there's no evidence of his name or a v5 build anywhere until his 453 which was only started a little before my comment. |
|
Here are the parameters, and the commit is definitely the one from v4.4.0 Perhaps by some odd co-incidence we started our jobs at the same time? I did start a second job 462... so everything seems to add up on my end |
|
Weird, at least the new build started. I'm going to blame this on some Jenkins failure. Here's what I think happened:
|
Notable changes: * child_process: “send()” now accepts an options parameter (cjihrig) nodejs#5283 - Currently the only option is “keepOpen”, which keeps the underlying socket open after the message is sent. * constants: “ENGINE_METHOD_RSA” is now correctly exposed (Sam Roberts) nodejs#5463 * Fixed two regressions which originated in v5.7.0: - http: Errors inside of http client callbacks now propagate correctly (Trevor Norris) nodejs#5591 - path: Fixed normalization of absolute paths (Evan Lucas) nodejs#5589 * repl: “start()” no longer requires an options parameter (cjihrig) nodejs#5388 * util: Improved “format()” performance 50-300% (Evan Lucas) nodejs#5360 PR-URL: nodejs#5559
|
All done! |
2016-03-08, Version 5.8.0 (Stable). @Fishrock123
Notable changes