<mrmargolis>
No luck. I'll try installing without rbenv
<headius>
ah
<headius>
you know what, that's an old version we removed some time ago
<headius>
we don't keep around snapshot versions for old releases very long
<headius>
what do you want to install? a lot has changed since 9.0.0.0
<mrmargolis>
I have 9.0.4.0 but it blows up when I try to use the redis gem. I wanted to try master.
<mrmargolis>
"Exception in thread "Ruby-0-Thread-11: /app/user/app/middleware/websockets.rb:15" java.nio.channels.ClosedSelectorException"
<mrmargolis>
Seems related to sockets inside of a Thread. Was going to try master and then see if I can create a minimal reproduction outside of this app.
lanceball is now known as lance|afk
pitr-ch has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<headius>
mrmargolis: ahh I see
<headius>
that comes up occasionally...if you can get a full backtrace that would help
<headius>
I'd open a bug now to start discussing it, and you can try nightly snapshot builds at ci.jruby.org...but I don't know how to get rbenv to install them
<headius>
looks like there were successful snapshot builds last night
<mrmargolis>
Thanks. I'll work on a reproduction and open up a bug if I can get a more detailed backtrace.
pitr-ch has joined #jruby
pitr-ch has quit [Client Quit]
<GitHub95>
[jruby] ylansegal opened issue #3594: Performance degradation in Rails App with 9.0.4 (Compared to 1.7.23) http://git.io/vzLMl
<GitHub169>
[jruby] ylansegal opened issue #3595: Performance degradation in Rails App with 9.0.4 (Compared to 1.7.23) http://git.io/vzLMD
pietr0 has quit [Quit: pietr0]
pitr-ch has joined #jruby
pitr-ch has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
e_dub has quit [Quit: ZZZzzz…]
edub has quit [Read error: Connection reset by peer]
bbrowning_away has quit [Ping timeout: 240 seconds]
bbrowning has joined #jruby
clayton has quit [Excess Flood]
clayton has joined #jruby
djellemah has quit [Ping timeout: 240 seconds]
djellemah has joined #jruby
mkristian_ has quit [Read error: Connection reset by peer]
mkristian has joined #jruby
kares has quit [Ping timeout: 240 seconds]
shellac has joined #jruby
yfeldblum has joined #jruby
drbobbeaty has quit [Ping timeout: 276 seconds]
benlovell has joined #jruby
anaeem1_ has joined #jruby
anaeem1_ has quit [Remote host closed the connection]
<lupine>
oh god, gserver. why?
mkristian_ has joined #jruby
<eregon>
headius: yeah I should document that more precisely
<eregon>
It's non trivial, and git subtree split sort of only half works unfortunately since it tries to reference some commit directly from ruby/spec at some point
mkristian has quit [Ping timeout: 276 seconds]
yfeldblum has quit [Ping timeout: 255 seconds]
brixen has quit [Ping timeout: 276 seconds]
brixen has joined #jruby
<GitHub73>
[jruby] chrisseaton pushed 1 new commit to truffle-head: http://git.io/vzYK9
<GitHub73>
jruby/truffle-head da784fd Chris Seaton: Merge branch 'master' into truffle-head
bbrowning is now known as bbrowning_away
<GitHub111>
[jruby] chrisseaton pushed 1 new commit to master: http://git.io/vzY6X
<GitHub111>
jruby/master 712f62c Chris Seaton: [Truffle] Support -Dgraal options.
<GitHub172>
[jruby] chrisseaton pushed 2 new commits to truffle-head: http://git.io/vzY6y
<GitHub172>
jruby/truffle-head 4c917bb Chris Seaton: Merge branch 'master' into truffle-head...
<GitHub172>
jruby/truffle-head d2beb01 Chris Seaton: [Truffle] Graal options are now -Dgraal. rather than -Dgraal.option.
yfeldblum has joined #jruby
drbobbeaty has joined #jruby
Tristitia has quit [Remote host closed the connection]
rsim has quit [Read error: Connection reset by peer]
rsim has joined #jruby
benlovell has quit [Ping timeout: 240 seconds]
<GitHub105>
[jruby] etehtsea opened pull request #3597: Add spec for SizedQueue#num_waiting regression (master...num_waiting_regression) http://git.io/vzYQN
<nirvdrum>
StringSupport.strLengthWithCodeRangeNonAsciiCompatible and StringSupport.codeRangeScanNonAsciiCompatible can return different code ranges for the same string.
benlovell has quit [Ping timeout: 264 seconds]
<nirvdrum>
I think it's only empty strings that end up problematic.
subbu has quit [Ping timeout: 246 seconds]
subbu has joined #jruby
thedarkone2 has joined #jruby
benlovell has joined #jruby
shellac has quit [Quit: Ex-Chat]
Tristitia has joined #jruby
<nirvdrum>
lopex: I don't suppose you're around, are you?
pitr-ch has joined #jruby
pitr-ch has quit [Client Quit]
benlovell has quit [Ping timeout: 255 seconds]
benlovell has joined #jruby
<lopex>
nirvdrum: hiya
<nirvdrum>
lopex: Hi there. I was going to ask about an issue with StringSupport.strLengthWithCodeRange, but it looks like that was ported from MRI.
<nirvdrum>
And it looks like MRI has the same problem I was going to ask about :-/
<nirvdrum>
I'd still appreciate it if you could double-check for me though. I may very well just be wrong.
<nirvdrum>
The issue seems to be that if a fixed-width encoding is used, this function does not in fact return the code range as it indicates.
cremes has quit [Ping timeout: 246 seconds]
<lopex>
nirvdrum: yeah, the fixed width doesnt
<lopex>
I'll look at mri
subbu has quit [Ping timeout: 240 seconds]
<lopex>
nirvdrum: I guess, your issue is that fixed width encodings cannot have truncated chars, but can have malformed
subbu has joined #jruby
<lopex>
well, multibyte ones can have truncated one too of course
<lopex>
but that's somewhat different
cremes has joined #jruby
<nirvdrum>
lopex: Well, it's a bit worse than that. US-ASCII is fixed width and can have only 7-bit chars.
<lopex>
right
<nirvdrum>
This method always sets it to CR_UNKNOWN.
<lopex>
yes
<nirvdrum>
I assume it's so the length can be calculated without a full scan.
<nirvdrum>
But that's what the non-with-cr method does.
<lopex>
right, so I guess we could add a fast lookup field to encoding that manifests whether there's malformed chars possible
<nirvdrum>
If I just remove that first case, I think the method works as I'd expect.
<lopex>
and another for truncated
<nirvdrum>
It does a full scan, but I'd expect that if I'm trying to figure out the code range anyway.
<lopex>
nirvdrum: yeah, it's a good find
<lopex>
but we still could optimize that the right way
<nirvdrum>
I can make the change here, but I'm worried about impacting JRuby.
<lopex>
perf wise ?
<nirvdrum>
Yeah. RubyString#strLength uses this method.
<nirvdrum>
I think it should only call this method if CR is CR_UNKNOWN, otherwise use StringSupport.strLength.
<nirvdrum>
And that looks like what MRI does.
<lopex>
you mean single_byte_optimizable ?
<lopex>
ah, the other one
<lopex>
right
<nirvdrum>
Right. The other one will do the fast length check (no need to do a full scan if you already know the CR).
<lopex>
nirvdrum: not sure if that packing was a good idea, just wanted to eliminate allocations
<nirvdrum>
lopex: It seems fine. Otherwise you're dealing with arrays.
<nirvdrum>
I really wish Java just had a tuple type.
<lopex>
of holders objects
<lopex>
there's a few in jcodings
<nirvdrum>
Nifty. I'll take a look.
<lopex>
nirvdrum: yeah, that string routines in mri are heinous
<nirvdrum>
What do you think of the other issue I reported about different code ranges being calculated?
<nirvdrum>
I think the one with the string length is the correct one.
<nirvdrum>
It reports "".force_encoding("utf-16le") as CR_7BIT.
<nirvdrum>
The general code range scan method reports it as CR_VALID.
<lopex>
yeah, mri seems to do the same
<lopex>
er wait
<lopex>
there's if (!*cr) *cr = ENC_CODERANGE_7BIT in both cases
<eregon>
you want documentation on my git magic hacks? :)
<headius>
I managed to find it by searching IRC logs :-D
<headius>
seems to have worked fine on ruby-2.3 branch
<eregon>
yeah, that is what I do
<headius>
I started trying to push too but it did scary looking things so I stopped it
<eregon>
but after getting the commits from jruby back to ruby/spec
<headius>
ahh I see
<eregon>
at least for the master branch
<eregon>
and that part is sort of broken with subtree, so it needs some more care
<headius>
hmmm so if there are unpushed commits on our side should I undo the pull and push first?
<eregon>
are there any new specs committed to the ruby-2.3 branch directly in jruby?
<headius>
hard to tell
<kares>
should be none - last I checked
<headius>
I haven't made any recently
<eregon>
seems two with git log -p origin/ruby-2.3 ^origin/master spec/ruby
<headius>
nice
<headius>
better to unpull and push or just duplicate them on ruby/spec?
<eregon>
if it's two we could just cherry-pick or sth simpler
<headius>
I'll do that today then
<eregon>
easiest is probably just create path for them, or copy the files if there is no other midification, and copy/paste the commit message and commit ot rubyspec
<headius>
enebo: anything you need from me for 1.7.24? I saw your comment on the stat bug
<eregon>
the merge after should be just fine
<headius>
yeah no problem
<eregon>
patch*
<enebo>
headius: well I think looking at io/console on windows could be good
<headius>
I'll get 'em in
<enebo>
headius: I might have the stat crasher fixed now too
<headius>
enebo: damn, I was hoping you'd say I'm good
<enebo>
headius: but I need to come up with some more tests in jnr-posix
<headius>
heheh
<enebo>
headius: hahah
<headius>
ok
<enebo>
headius: also look at what we have marked in case we have anything embarrasing in there
<headius>
so I'll finish updating specs on ruby-2.3 and ruby/spec and then have a look at that
<headius>
ok
<eregon>
I need to go, I'll try to document on the ruby/spec wiki next time I do the merge
<enebo>
lunch
pawnbox_ has joined #jruby
pawnbox has quit [Ping timeout: 246 seconds]
<headius>
eregon: thank you
DomKM has joined #jruby
bb010g has joined #jruby
bbrowning is now known as bbrowning_away
blaxter has quit [Ping timeout: 240 seconds]
lanceball is now known as lance|afk
lance|afk is now known as lanceball
<lopex>
nirvdrum: I meant the other issue wrt that 7BIT and rb_enc_strlen_cr
<lopex>
nirvdrum: and you're right it becomes 7BIT for nonascii since it doesnt enter that loop for empty string
<nirvdrum>
lopex: I spoke with eregon for a while about it. It looks like it will return CR_UNKNOWN, but MRI expects this. I.e., it's not a bug, just a very questionable design decision.
<lopex>
hmm, that appears to be missing in jruby repo
<headius>
hah, jruby-cdc
<headius>
we should probably nuke that repo
<enebo>
hahaha
<enebo>
lopex: nice
<enebo>
lopex: I totally believe this is your first favorite behavior bug :)
<lopex>
enebo: at that point mri was secured against that
<lopex>
enebo: jruby was also never compatible wrt ivars for fixnums nd symbols
<lopex>
but now since those are frozen it's good
<lopex>
enebo: btw, did you have any good beers in Japan ?
<enebo>
lopex: yeah plenty of them
<lopex>
i had of couple of US ones for christmas
<enebo>
lopex: I think they have some devent breweries. I think US is still hard to compete with from the best of the best perspective since there are >4000 breweries in US now
<lopex>
enebo: like those from EU ?
<enebo>
lopex: just a lot more experimentation
<lopex>
US beers are freaking expensive here
<enebo>
lopex: EU is a tougher comparison because there are many old breweries who make the same beer for centuries
<enebo>
lopex: so they are good beers but I am more excited by new brewing at this point
<enebo>
lopex: but EU has plenty of new brewries too just not as much
<enebo>
lopex: but lots of great traditional beers and some innovative new brewers makes EU more complicated to compare with
<lopex>
enebo: yeah, the whole EU is too
<enebo>
lopex: yeah and I am lumping all of EU together but if you go to France you will not find a lot of beer
<lopex>
enebo: de molen are the best UE onces for me
<lopex>
*ones
<enebo>
lopex: so it is weirdly spotty
<enebo>
lopex: yeah I bought some de moel in Japan since they hardly distribute in the US
<lopex>
though the rest of belgian styles not so much
<enebo>
lopex: de molen is from netherlands isnit it?
<lopex>
you got me
<lopex>
let me reassure
<enebo>
lopex: yeah netherlands
<lopex>
yeah
<lopex>
enebo: I feel stupid now
<lopex>
well, above level of me feeling stupid
<headius>
EU is a tougher comparison because there's like 4000 breweries that just do lagers
<enebo>
headius: though in germany there are like 200 lagers all with different brands and nearly all are now made by one company in the same factory
<lopex>
headius: yeah, I guess they brew mostly lagers in Czech Republic for example
<enebo>
headius: makes you wonder if they might accidentally slip wrong recipe into the wrong beer
<enebo>
but perhaps their consolidation is not much different than US and inbev
<enebo>
and I doubt what I said is all that true…I just know there is one brewing location which brews several different label beers in DE (although it is just contract brewing I guess — by that logic I should be hard on 21st amendment for being made by cold spring brewery)
<headius>
enebo: I wouldn't be surprised if most of the german beers are just a couple companies now
<enebo>
It is one country where a lot of consolidation has been happening over decades but other than craft same is true of US
<headius>
did SABMiller + ABInBev happen yet?
<enebo>
don’t know
<kares>
not the case in Czech - biggest breweries are in different cities -> different factories :)
<kares>
and there's like a very lot of small ones (compared to population)
<kares>
not that I know much about beer but I live here :) !
<enebo>
kares: it is also possible it is still true in DE too since it may just be consolidated ownership
<enebo>
kares: but Buvar and Jupiler and many of the bigger Czech beers are owned by same entity right?
<kares>
was suprising to hear for me that its done from one place in DE
<kares>
enebo: Budvar and Plzen are the biggest
<kares>
Budvar still state owned
<kares>
partially - the last one :)
<enebo>
Someone was talking about how all berliner weiss are made in the same factory
<enebo>
and not actually a beliner weiss but that is a different issue :)
<enebo>
but lots of contract brewing happens now around the world
<kares>
Plzen owns other labels - but everyone nows that :)
<enebo>
:)
<kares>
enebo: that's true - actually there's this one guy here who just makes beer where ever they allow him and rarely does the same twice
<enebo>
I did not know about Budvar being state-owned…I wish they would have won their lawsuit in US against budweiser
<kares>
best beers I had :)
<kares>
yes Budweiser is not Budvar
<enebo>
but they took budweirser to court over name and US threw it out
<lopex>
kares: we always went to Czech for beers in the past (especially from southern Poland)
<kares>
but the fact is you only get true Budvar (exceptional taste) in like one spot in Prague and than in Budejovice where they make it
<enebo>
sounds like most countries rule in favor of budvar
<lopex>
kares: but there's a beer revolution in Poland too
<kares>
lopex: still pendin my visits to Polland - but I heard you got some great wine :)
<enebo>
lopex: yeah it was crazy in 2014. Do you think it has grown recently?
<kares>
my father is Hungarian and there's a saying -> Polish and Hungarian are too good friend whom drink each other's wine
<kares>
* two
<enebo>
lopex: but weirdly in 2014 there was only like three bars which carried them in Krakow
<lopex>
kares: I'm not into wines, I dont even know that polish ones are known
<lopex>
enebo: it's a little bit more silent now
<kares>
lopex: so what would be a good beer to try in Krakow?
pawnbox has joined #jruby
<lopex>
kares: craft beers ?
<kares>
yy
<kares>
local ones
<lopex>
kares: I dont know since there's o lot of change now, I'd have to ask (I live in katowice which is about 60km from krakow)
<lopex>
the name is taken literally from "empire strikes back" movie title polish translation
<kares>
oh wow quite a strong one
<kares>
lopex: can understand that much polish :)
<enebo>
lagering
<enebo>
hmm seems description may have some mistakes
<lopex>
kares: that was for enebo :)
<enebo>
unless people refer to secondary fermentation as lagering for ales
<lopex>
kares: so he can feel like a foreigner for a while
<kares>
:)
<kares>
lopex: which part of Poland are you actually from?
<enebo>
lopex: I always feel like a foreigner (not the band)
<lopex>
kares: I guess I heavent heard much about craft beers from other parts of poland
<lopex>
kares: south, near krakow
yfeldblum has joined #jruby
<kares>
lopex: its going to be a small brewery - this pub specializes for only having those on tap
<kares>
interestingly I never noticed a polish beer on tap ... up till now so I'll probably go for a try tomorrow
<lopex>
kares: ah, yeah, but it's one of the largest among craft ones here
<lopex>
they all have about 10% of the market
<kares>
najs
<lopex>
so 90% is international corporate lager :)
<enebo>
kares: so jruby-openssl is also updated in 1.7 branch right?
<kares>
but they're still called underground :)
<lopex>
but it jumped from 2-3% in last two years
<kares>
enebo: yep - to 0.9.13 - that is intentional to no go to 0.9.14 up till next
<enebo>
ok
<enebo>
kares: but it will be a bullet point in release notes
<kares>
ok
<enebo>
I think I have both windows stat issues fixed so we can probably move to some verification to get 1.7.24 out this week (and hopefully 9.0.5.0 too)
<kares>
yay!
<kares>
after 9.1 will there be 9.0 releases?
<enebo>
kares: no
<enebo>
THERE CAN BE ONLY ONE
<enebo>
:)
<lopex>
headius: the unix socket leak I had was on 1.7
<kares>
enebo: np for me :)
<headius>
lopex: gah
<enebo>
kares: at this point the changes between points have such little backward compat breaking material we will be trying to just roll with this strategy
<headius>
well then it isn't jruby or there's two leaks
<enebo>
headius: you got 8 fingers and two thumbs…plug it up
<headius>
yeah yeah
subbu is now known as subbu|lunch
<headius>
1.7 didn't show the leak I found
<enebo>
ok new jnr-posix coming….yeee haw
<headius>
but it needs to be fixed for next 9k
<enebo>
headius: you talking to me or lopex?
<enebo>
:)
<enebo>
you just mean socket leat on 9k still needs fixing
<enebo>
LEAT
<enebo>
People should be able to tell by my typos when I am on a KVM in the basement
<lopex>
headius: yeah, I'm still not sure it's a jruby, I had to reboot since that app is used daily
<lopex>
so no other info except kernel oom
<lopex>
will try to test that on another
<headius>
well you should be able to see a jruby leak pretty easily
<headius>
if it's leaking something native it won't JVM OOM
<lopex>
yeah, it's not jvm ooming
<GitHub134>
[jruby] enebo pushed 1 new commit to jruby-1_7: http://git.io/vzG27
<GitHub134>
jruby/jruby-1_7 ee09c3c Thomas E. Enebo: Upodate to latest jnr-posix. Fixes #3525. stat.writable? incorrectly reporting false for some directories on Windows 7
<enebo>
gah
bb010g has quit [Quit: Connection closed for inactivity]