-
Notifications
You must be signed in to change notification settings - Fork 93
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
update to libv8 6.0 #65
Comments
I'm getting
on is my platform built @ignisf or am I doing something else wrong? |
looks like this issue is #66 and I think related to the --start-group linker error (unknown option) |
haha thanks @ignisf -- can't tell if this is introducing the issue or fixing it? Assuming introduces, but don't understand how this is happening on mini_racer mainline (through bundler) when dependency is @ 5.9 |
I'll revert this commit and release another beta. Bundler is not selecting the correct version of libv8 I think. Try uninstalling 6.*. Sorry for taking so long to respond. My ISP has a broken optical connection. |
it's all good @ignisf! Thanks for the response. |
You were right-- uninstalling 6.0 got the correct libv8 version to load up, and has isolated the issue to 6.0beta. Thanks for the help @ignisf :) |
6.0 beta2 is up |
Everything works @ignisf thanks! |
Tests all check out. :) |
6.0 released 🎉 |
@seanmakesgames are you testing it on production again? :D |
That's just how he roll :) |
Can you see significant performance improvement? |
Haha, nah-- was running against my test suite. I'll be 'testing on production' once mini_racer has upped its version number. ;) I didn't do any benchmarking as part of my test cases because I run them multicore. It did 'feel' faster, but that's not useful information. :P |
Haha. When I updated Node to 8.4.0 (with v8 6.0) it was about 40% faster. |
That's a question for @SamSaffron (whom I'm sure is busy atm, or he'd probably have said something already) ;) |
@krzysiek1507 I don't have an eta, but once @seanmakesgames is happy that 6.0 is stable I am happy to bounce the version. My initial testing definitely did not show a 40% improvement in my microbench maybe 5-10% |
Note, I am travelling over the next 3 weeks so I am not 100% sure I will be super fast here. |
Thanks, guys! You are doing great work! |
@SamSaffron 6.0 appears stable. I'm good with a rev bump. |
When I installed it from my repo I saw error:
Version: |
Any update on this? |
Will do another release next week |
Thanks! |
confirmed @ignisf getting: /home/sam/.rbenv/versions/2.3.1/bin/ruby: symbol lookup error: /home/sam/Source/mini_racer/lib/mini_racer_extension.so: undefined symbol: _ZN2v84base5debug27EnableInProcessStackDumpingEv on GCC with 6.0.286 , can you have a look? |
morning, back from vacation, will try to check it out |
cool, thanks!
…On Mon, Sep 25, 2017 at 5:57 PM, Petko Bordjukov ***@***.***> wrote:
morning, back from vacation, will try to check it out
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#65 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAAUXSX6oyCLRm5DTEelNxDxhjnk_oNKks5sl1zZgaJpZM4PBotX>
.
|
the missing symbol was due to the reordering of the libraries. Who knew Google's wiki would lie... Should be fixed in master and the 6.0 branches |
just an FYI, we will be losing FreeBSD support with this version due to lack of upstream support. I decided to not hold up the release more. Will hopefully work out a solution and release binaries at a later stage. |
Fine by me. ;) |
OK, my plan for tomorrow is to bench it on Discourse! |
This is not great... my bench suite for discourse is hanging ....
…On Thu, Dec 14, 2017 at 6:06 AM, seanmakesgames ***@***.***> wrote:
Fine by me. ;)
I'm excited to see how this release performs in the live environment. 😄
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#65 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAAUXd2_6bZJAtx988KhpD3DbVziT7SIks5tACAxgaJpZM4PBotX>
.
|
Ouch ... 6.0 is hanging as well, this is looking like its probably a bug in
mini_racer.
…On Fri, Dec 15, 2017 at 12:45 PM, Sam Saffron ***@***.***> wrote:
This is not great... my bench suite for discourse is hanging ....
On Thu, Dec 14, 2017 at 6:06 AM, seanmakesgames ***@***.***>
wrote:
> Fine by me. ;)
> I'm excited to see how this release performs in the live environment. 😄
>
> —
> You are receiving this because you were mentioned.
> Reply to this email directly, view it on GitHub
> <#65 (comment)>,
> or mute the thread
> <https://github.com/notifications/unsubscribe-auth/AAAUXd2_6bZJAtx988KhpD3DbVziT7SIks5tACAxgaJpZM4PBotX>
> .
>
|
That's strange. My full test suite passed. :\ |
oops @ignisf looks like 6.2 is hanging for me during script runs. I will try to build a repro for this. I wonder if this was a temporary bug in v8 or if it is fixed in 6.3 6.0 seems to work fine though. |
the repro of the bug is to run: https://github.com/discourse/discourse/blob/master/script/benchmarks/markdown/bench.rb but that involves a full discourse install... |
will look into building betas for 6.1 and 6.3 |
@SamSaffron, just released 6.3 beta. |
Oh no :( looks like this is hanging as well, going to create a simple repro script here and ask v8 folks for help. |
@SamSaffron, are you sure you're testing with mini_racer linked against the 6.3 beta and not by any chance linked against 6.2? |
yes, 100% 6.3.292.48.0beta1 in my gemfile lock ... trying to build a repro |
I am pretty sure this is the bug:
https://bugs.chromium.org/p/v8/issues/detail?id=6933
Can you compile 6.4 to confirm it is fixed there? This kind of sucks cause
it means we can not upgrade till 6.4 is released OR unless we backport it.
…On Tue, Dec 19, 2017 at 8:32 AM, Petko Bordjukov ***@***.***> wrote:
@SamSaffron <https://github.com/samsaffron>, are you sure you're testing
with mini_racer linked against the 6.3 beta and not by any chance linked
against 6.2?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#65 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAAUXazuFYVy2lZi_fU1mBOLvN1ZCaN0ks5tBtnsgaJpZM4PBotX>
.
|
OK, v8 team say it backports very cleanly, can you try maybe adding that patch and doing another 6.3 release? |
Hey, sounds good, will look into backporting. |
Will cut another release for you to test tomorrow if the CI passes tonight :) |
Awesome! Thanks for that!
…On Tue, Dec 19, 2017 at 10:25 AM, Petko Bordjukov ***@***.***> wrote:
***@***.***
<rubyjs/libv8@5777dbd>
Will cut another release tomorrow if the CI passes tonight :)
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#65 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAAUXTKAHYdHtq0ljKb2IXY0GGLdE8Gbks5tBvSBgaJpZM4PBotX>
.
|
@SamSaffron let me know when you've tested the patch on this. I'll run my suite against it after. |
Looking good on my side! @ignisf bench is no longer hanging, 6.3 appears slower than 5.9 in some of the bench suite (and faster in other bits): 5.9
6.3
Benchmark is: https://github.com/discourse/discourse/blob/master/script/benchmarks/markdown/bench.rb |
.0 and .1 releases are up on rubygems. |
prepping to run the suite on this (have been making changes, so want to make sure tests are passing first) ;) |
everything looks good on my end. tests pass. no funky behavior (more than usual) :P |
It is holiday season, I will try to get a new stable out next week, but will only upgrade Discourse the week after |
thanks heaps @ignisf for the work on libv8 6.3 we now have a new release. |
What about 6.4? They promise a significant optimization. |
I too am eager to push libv8 6.4. Just have to find some time to do so. |
am running my test suite against 6.0.286.54.0beta1
wanted a place to write down the results. :)
@SamSaffron @ignisf
The text was updated successfully, but these errors were encountered: