deps: update libuv to 0.10.36

[saghul]

R=@tjfontaine

[misterdjules]

Thanks @saghul!

@tjfontaine Tests results for Unices: http://jenkins.nodejs.org/job/node-review-unix-julien/42/ and Windows: http://jenkins.nodejs.org/job/node-review-windows-julien/34/.

So far, I've seen two suspicious failures, only on SmartOS: http://jenkins.nodejs.org/job/node-review-unix-julien/42/DESTCPU=ia32,label=smartos/tapTestReport/simple.tap-30/ and http://jenkins.nodejs.org/job/node-review-unix-julien/42/DESTCPU=x64,label=smartos/tapTestReport/simple.tap-473/.

[saghul]

Make that 0.10.35 :-)

simple/test-http-many-keep-alive-connections seems to consistently fail on my machine, not sure about the upgrade is the culprit though.

[misterdjules]

@saghul Thank you!

As far as I know, simple/test-http-many-keep-alive-connections failing is expected. However, test-child-process-fork-net2.js and test-signal-unregister.js failing is new to me.

[saghul]

Force pushed updating to 0.10.36. There was a problem with 0.10.35 which was caught by iojs.

[saghul]

@misterdjules can you please re-run the CI? Locally (Linux) I only get the expected failure:

git|uv_01034)saghul@alpha:~/src/node
$ make test-simple 
make -C out BUILDTYPE=Release V=1
make[1]: Entering directory '/home/saghul/src/node/out'
make[1]: Nothing to be done for 'all'.
make[1]: Leaving directory '/home/saghul/src/node/out'
ln -fs out/Release/node node
/usr/bin/python tools/test.py simple
=== release test-http-many-keep-alive-connections ===                          
Path: simple/test-http-many-keep-alive-connections
Command: out/Release/node /home/saghul/src/node/test/simple/test-http-many-keep-alive-connections.js
--- TIMEOUT ---
[02:34|% 100|+ 621|-   1]: Done                                                
Makefile:114: recipe for target 'test-simple' failed
make: *** [test-simple] Error 1

[misterdjules]

@saghul Building/testing here: http://jenkins.nodejs.org/job/node-review-julien/31/default/.

[misterdjules]

@saghul Builds results for UNICEs here: http://jenkins.nodejs.org/job/node-review-unix-julien/ and for Windows here: http://jenkins.nodejs.org/job/node-review-windows-julien/.

The only tests failing that I'm not sure are expected to fail are:

• test-net-error-twice.js on SmartOS.
• test-tls-session-cache.js on Linux.

The rest of the failures look to me like they're expected.

@joyent/node-coreteam Thoughts?

[tjfontaine]

I believe (unfortunately) that tls-session-cache is an expected failure currently, but deserves a second look, and can probably be verified by looking at the test results for v0.10.36-release.

net-error-twice seems like a behavior change on the order on which we expect events to happen, and does seem like a regression in behavior which should be further investigated, though the test itself does seem like it is slightly non-deterministic.

[misterdjules]

@saghul @joyent/node-coreteam test-tls-session-cache.js failing on Linux has happened before, and it seems it should be put in the same category of "spurious failures".

I went back in the history of the nodejs-v0.10 Jenkins job, and haven't found yet any occurence of test-net-error-twice.js failing on SmartOS. The only "expected spurious failure" I found on SmartOS so far is test-child-process-fork-net2.js.

[trevnorris]

So is there still a problem with test-net-error-twice?

[misterdjules]

@trevnorris I haven't had the time to investigate, I'll put that at the top of my priority list and hopefully I can find some time to work on it soon.

[saghul]

FWIW, I don't see how that test failing can be a libuv issue :-S

[trevnorris]

@saghul Same here. Don't see how the two are related.

[misterdjules]

@saghul @trevnorris Looking at it now.

[misterdjules]

@saghul @trevnorris Indeed, test-net-error-twice.js' failure is not due to this libuv upgrade. I'll create a separate issue for this but it shouldn't prevent landing this change.

[misterdjules]

Just FYI, created #9325 about test-net-error-twice.js failing.

[misterdjules]

@saghul @trevnorris I'll try to do an actual code review tomorrow (Thursday) now that we know that no regression has been introduced by this upgrade.

[misterdjules]

I just want to make sure we acknowledge that this doesn't change the ABI because sizeof(AFD_POLL_INFO) >= sizeof(AFD_POLL_INFO*).

[orangemocha]

+1 it looks ok.

[misterdjules]

@saghul @trevnorris @tjfontaine See my comments inline, but otherwise LGTM.

[saghul]

No, it doesn't in this case because its a union between the type and a
pointer to it.
On Mar 5, 2015 9:07 PM, "Julien Gilli" [email protected] wrote:

@saghul https://github.com/saghul @trevnorris
https://github.com/trevnorris @tjfontaine
https://github.com/tjfontaine See my comments inline, but otherwise
LGTM.

—
Reply to this email directly or view it on GitHub
#9274 (comment).

[misterdjules]

@saghul Ok thanks, we agree on that then 👍

[misterdjules]

@saghul Did you see this comment too?

[saghul]

@misterdjules sorry, missed that one. Just replied.

[misterdjules]

@saghul Thank you! @joyent/node-coreteam Landing asap.

[misterdjules]

Thank you again @saghul, landed in dcff5d5!

[saghul]

Great! \o/
On Mar 6, 2015 8:44 PM, "Julien Gilli" [email protected] wrote:

Thank you again @saghul https://github.com/saghul, landed in dcff5d5
dcff5d5
!

—
Reply to this email directly or view it on GitHub
#9274 (comment).