pitr-ch has quit [Read error: Connection reset by peer]
pitr-ch has joined #jruby
temporalfox has quit [Remote host closed the connection]
temporalfox has joined #jruby
brauliobo_ has joined #jruby
brauliobo has quit [Read error: Connection reset by peer]
lopex has joined #jruby
mkristian has joined #jruby
<GitHub140>
[jruby] pitr-ch commented on commit 43c8e4b: Yeah, I should have added a TODO at least to track it. The `@Volatile` should not leak here. http://git.io/vBTD7
pitr-ch has quit [Read error: Connection reset by peer]
pitr-ch has joined #jruby
pitr-ch has quit [Read error: Connection reset by peer]
ebarrett_ is now known as ebarrett
pitr-ch has joined #jruby
pitr-ch has quit [Read error: Connection reset by peer]
ruurd has joined #jruby
pitr-ch has joined #jruby
pitr-ch_ has joined #jruby
pitr-ch has quit [Read error: Connection reset by peer]
pitr-ch_ has quit [Read error: Connection reset by peer]
pitr-ch has joined #jruby
vtunka has joined #jruby
pitr-ch has quit [Read error: Connection reset by peer]
blaxter has joined #jruby
shellac has joined #jruby
pitr-ch has joined #jruby
mdedetrich has joined #jruby
mdedetrich has quit [Client Quit]
<GitHub43>
[jruby] mkristian commented on commit 5a7bd3f: the thing here is that "uri:classloader://META-INF/jruby.home" is just treated as a filesystem path and normalizing means to replace "//" with "/"... http://git.io/vBTde
waka has quit [Remote host closed the connection]
<GitHub19>
[jruby] mkristian opened issue #3487: NormalizedFile should keep the uri part of the path as is http://git.io/vBTdn
mdedetrich has joined #jruby
waka has joined #jruby
pitr-ch has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<GitHub73>
[jruby] kares created force-backtrace-nil (+4 new commits): http://git.io/vBTFg
<GitHub73>
jruby/force-backtrace-nil 65145de kares: test how Ruby exceptions are behaving on the Java API side
<GitHub73>
jruby/force-backtrace-nil eb2aa90 kares: allow RaiseException() with passed nil backtrace to correctly force the backtrace as nil
<GitHub73>
jruby/force-backtrace-nil 7841a87 kares: improve forceBacktrace (called on RaiseException()) to keep backtrace() lazy...
<GitHub108>
[jruby] kares commented on commit 5a7bd3f: thought so, thanks. just wanted to make sure I do not break anything - build should be :green_heart: :tea: http://git.io/vBkzS
<chrisseaton>
I'm not seeing that further down the pipeline
<chrisseaton>
Maybe it's not our fault...
<chrisseaton>
I mean maybe it is our fault
<enebo>
chrisseaton: this commit c293b6310122c
<enebo>
chrisseaton: If anything I think it might mark more things as isnewline but it was mostly removing lots of extra “is this a newline…delete code”
<enebo>
chrisseaton: but I also did change BodyTranslator in that commit
<enebo>
chrisseaton: So maybe I messed something up there?
<headius>
enebo: simple example above seems to parse right
<enebo>
mkristian: all my m2/repository/rubygems/* files are HTML with a 301 moved error in them on windows
camlow325 has quit []
<enebo>
mkristian: I can try on non-windows to see
<mkristian>
enebo, oh - that I think debugging maven now is another huge thing now :)
<enebo>
mkristian: DOES NOT COMPUTE :)
<mkristian>
enebo, will not look into windows and maven now
<enebo>
mkristian: I will nuke rubygems org on macos and make sure it is also a windows-only thing
camlow325 has joined #jruby
<mkristian>
enebo, just keep me in the loop. I never tried to build on windows as I just mounted my macos path inside the windows VM
<enebo>
mkristian: I have not been building 1.7 on windows but have been building master there last week
<mkristian>
ok
<enebo>
mkristian: seems to have compiled fine on MacOS
<enebo>
mkristian: I rm’d all the rubygems maven artifacts before I did a clean build so it appear to have downloaded them fine
<headius>
mkristian: I was able to run enebo's script on my windows VM without issues
<headius>
so there may be something environmental
<headius>
I'm following his process exactly now to confirm
<enebo>
I did not paste top of build but this shows windows not following permanent refdirect for some reason
<headius>
I'm stlll downloading the internet
<enebo>
It just saves the HTML response which comes along with the HTTP 301 error
<headius>
what mvn version?
<enebo>
neat if you curl these urls it will not follow by default
<headius>
I'm running 3.3.3
<enebo>
headius: well I am not running mvnw on 1.7 branch
<headius>
neither am I
<enebo>
ok so mine is pre 3.3
<headius>
could be related
<enebo>
ok yeah could be. I can try getting a newer version
<enebo>
I just was building master and using ./mvnw
<headius>
build completed fine for me
<enebo>
so it could just be maven is loading an older artifact which is not hip to redirects and jar-deps is using it?
<mkristian>
enebo, I never added mvnw to jruby-17 branch
<enebo>
mkristian: yeah I thought not
<enebo>
mkristian: ubt could my mvn be too old somehow?
<headius>
we ought to do that just so maven isn't an a required step to build 1.7
<headius>
downloading maven
<enebo>
I am using 3.0.2 on windows for 1.7 build
<headius>
wow
<headius>
WOW
<enebo>
I can update to 3.3 if need be (I was lazy because it is manual on windows)
<enebo>
but we are not using any polyglot features or anything on 1.7
<headius>
dude, that's four years old :-)
<subbu>
headius, enebo over the next several weeks, I'll start implementing additional IR instrs to move more of the block-call-protocol-stuff out of the runtime and bypass the whole complex logic in BlockBody and child classes.
<headius>
almost 5
<enebo>
subbu: excellent
<subbu>
it is a bit messy because of all the ruby core classes that call block code directly .. so, it will be hacky initially.
<headius>
subbu: that's excellent
<enebo>
headius: strange…I do not even think this machine is 5 years old
<headius>
enebo: yeah that's weird
<mkristian>
enebo, on 17 branch just needs maven-3 but 3.0.2 is kind of old
<headius>
so some time after 3.0.2 it probably started following redirects for artifacts
<subbu>
but, my hope is that we can get rid of BlockBody uses for the common case as this gets more refined.
<subbu>
so, Block with just be Block where yields will just call through a thin layer in Block and back into interp / JIT.
<headius>
subbu: that would be wonderful...then we'll be closer to having blocks inline at JVM level too
<enebo>
give me a couple of minutes to update and I will verify it is in fact my crusty maven
<subbu>
headius, right, that is my intention
<headius>
for cases where we can't or don't inline at IR level
<subbu>
i had sent notes back in May to you guys and i've started refining it last day or two and been staring at the code a bit more .. my memory is a bit rusty after having not looked at this code for a while now.
<headius>
enebo: script worked fine from a "space path"
<headius>
jruby is not installed in spaces
<subbu>
but, looks like it is doable .. with hacks initially, but can be cleaned up after.
<subbu>
that is the end of my PSA. you can continue with regularly scheduled programming.
<headius>
subbu: cool, thanks for the update :-)
<headius>
I will be in SE Asia for the next couple weeks but I'll be working
<chrisseaton>
enebo: sorry I think this newline thing is our fault
josh-k_ has quit [Remote host closed the connection]
camlow325 has quit [Read error: Connection reset by peer]
camlow325 has joined #jruby
<enebo>
chrisseaton: np. I totally forgot I changed it so a refresher is nice :)
<headius>
enebo: confirmed in cmd too
<headius>
ship it
<enebo>
rebuilding with newer maven
lanceball is now known as lance|afk
<enebo>
hmmm I suppose I should have deleted all of those corrupted files before I tried this
<enebo>
I was so lazy because I added maven to my windows env vars and it had a version specific path
<enebo>
did nirvdrum fall off the face of the planet?
<nirvdrum>
Yo.
<nirvdrum>
Just working on my presentation.
<enebo>
I had no reason to mention your nick other than missing your presence
<nirvdrum>
I committed some stuff last night :-)
<enebo>
GOOD NEWS EVERYONE…updating to a recent maven solved my problems
<headius>
huzzah!
<enebo>
replacing jruby.jar in my test install to see if the script also works for me
<enebo>
people I think really hate this io/console warning
mattwildig has joined #jruby
samphippen has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<enebo>
ok so I think this script is working for me. It does from my dev env anyways. If I copy over my jruby.jar into my last install bits it fails with the readline not found but I think this was a build error
<enebo>
so now I will regenerate dist bits and try again…hopefully we are golden this time around
mattwildig has quit [Ping timeout: 240 seconds]
<headius>
"This copy of Windows is not genuine."
<mkristian>
enebo, did you use the latest code from github ?
<enebo>
perhaps it is my age but I only associate genuine with leather
<enebo>
mkristian: HEAD of jruby-1_7 but I just copied the jar into last friday’s build
<enebo>
mkristian: so I will make full build + installer and verify it is fixed
pitr-ch has joined #jruby
<enebo>
mkristian: my local dev build on windows worked fine
<mkristian>
enebo, let me see if my test still fails - did not try after I backportet the last paths patches from master
<enebo>
mkristian: yeah good call. I am hoping that is fixed or it is a different bug
<enebo>
mkristian: it must be though
<enebo>
mkristian: It nearly exactly what my script did but I was using system and not ``
mattwildig has joined #jruby
<mkristian>
enebo, so my example does not work on 1.7.22 as well and it comes from my initial trial to create a new rails application on windows
<mkristian>
where I did find those `` in rails somewhere
<enebo>
mkristian: ok so if not a regression we just have something to fix
<enebo>
mkristian: but not for 1.7.23 unless it was something really basic in Rails you tripped over
<enebo>
mkristian: I run scaffold/mkgrations/new/server in my tests
<mkristian>
enebo, I ran `rails new something` and bundle install goes via these backticks
josh-k has joined #jruby
<enebo>
mkristian: rails new will run bundler successfully for me on windows
<enebo>
ok generating installer now getting closer to a real test on windows
<mkristian>
enebo, I am just trying gem install rails; rails new something; let's see
<enebo>
mkristian: I do this for full dist build; mvn clean deploy -Psonatype-oss-release -Prelease
<enebo>
mkristian: technically it is an empty never build shallow clone so the clean is not needed but I do it anyways
colinsurprenant has quit [Quit: colinsurprenant]
<mkristian>
enebo, so it is still a problem with me on windows-8.1 (or so) and rails-4.2.5
<enebo>
mkristian: I am nearly running my script so I will confirm or deny something with build being an issue or not
<mkristian>
ok
<enebo>
mkristian: the runner script was able to run through a rails session for me
<enebo>
mkristian: so you specifically cannot ‘rails new something’?
pitr-ch has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<mkristian>
enebo, same thing I put into this issue of mine
<GitHub110>
[jruby] headius pushed 2 new commits to ruby-2.3: http://git.io/vBtmT
<GitHub110>
jruby/ruby-2.3 0c93f88 Stefan Huber: Enumerable#chunk and Enumerable#slice_before no longer takes the initial_state argument [Feature #10958]
<GitHub110>
jruby/ruby-2.3 bb8807f Charles Oliver Nutter: Merge pull request #3490 from MSNexploder/feature_10958...
colinsurprenant has joined #jruby
<headius>
that's unexpected
<headius>
oh, kares excluded it...neat
<headius>
I will dig deeper
<GitHub152>
[jruby] chrisseaton pushed 2 new commits to truffle-head: http://git.io/vBtYG
<GitHub152>
jruby/truffle-head ae20878 Chris Seaton: [Truffle] Made the distinction between Truffle nodes and JRuby nodes very clear in the translator.
<GitHub152>
jruby/truffle-head e19a173 Chris Seaton: [Truffle] Copy newline state when desugaring.
<headius>
hmmm
<headius>
a["abc"] and b["abc"] dedup abc for Hash?
<headius>
perhaps because it's a literal key?
pitr-ch_ has joined #jruby
pitr-ch has quit [Read error: No route to host]
<headius>
mmm yup
<headius>
MRI treats the string in a["abc"] as fstring internally unless it turns out a is not a hash at call time
<headius>
in which case it "resurrects" the string as a new object
blaxter has quit [Ping timeout: 272 seconds]
<headius>
enebo: there's a handful of these optz
<headius>
in insns.def
lance|afk is now known as lanceball
griest has joined #jruby
<griest>
any chance you guys will add support for javafx 8?
<GitHub174>
[jruby] headius opened issue #3491: Frozen string optimizations from MRI http://git.io/vBtZh
<headius>
griest: I think JRubyFX is seeking contributors, but I don't see why it wouldn't support latest if we can get to it
colinsurprenant has quit [Quit: colinsurprenant]
colinsurprenant has joined #jruby
<griest>
trying to use javafx 8 with pure jruby causes problems
<griest>
java.lang.RuntimeException: Error: class sun.reflect.NativeMethodAccessorImpl is not a subclass of javafx.application.Application
<griest>
i think this is because jruby is trying to coerce some types or something
<griest>
that is caused by just trying to create a main class that extends the Javafx Application class
<griest>
To clarify: I have created my main class and then call Main.launch. the start method just contains some trivial code for a hello world program.
mkristian has quit [Quit: This computer has gone to sleep]
<headius>
ahhh
<headius>
yeah extending in JRuby has to do a lot of squirrely tricks
<headius>
I'd recommend you look at JRubyFX, which works around most of those issues
<GitHub117>
[jruby] pitr-ch commented on commit 01476b2: Yeah, it directly uses the gem configuration and local config can be used to override. I think that's simpler to manage in our volume of tested gems. http://git.io/vBqfH
waka has quit [Remote host closed the connection]
<griest>
headius: Do you know who is mostly "in charge" of the jrubyfx repo that I can bounce some ideas off of?
<headius>
enebo and nahi I believe (@tom_enebo and @nahi on twitter)
<GitHub2>
[jruby] chrisseaton pushed 1 new commit to truffle-head: http://git.io/vBqPX
<GitHub2>
jruby/truffle-head f24e0ac Chris Seaton: [Truffle] Workaround problem with instrumentation order for duplicated block bodies, in order to fix coverage.
<GitHub40>
[jruby] nirvdrum pushed 1 new commit to master: http://git.io/vBqNb
<GitHub40>
jruby/master 5446cc2 Kevin Menard: [Truffle] Fixed an NPE when a SharedMethodInfo has a null argumentsDescriptors.
enebo has joined #jruby
<GitHub54>
[jruby] nirvdrum commented on commit 5446cc2: @eregon @chrisseaton I don't know if storing `null` originally was done as a memory/allocation saving feature. If not, I can change the code to store the empty array in the `SharedMethodInfo` constructor and remove this conditional.... http://git.io/vBqAC
<GitHub53>
[jruby] chrisseaton pushed 1 new commit to truffle-head: http://git.io/vBqAr
<GitHub53>
jruby/truffle-head 9f7fefc Chris Seaton: [Truffle] Newlines on global assign.
<GitHub130>
[jruby] chrisseaton commented on commit 5446cc2: Yeah simpler is better - allocation the array in the constructor. http://git.io/vBqxx
enebo has quit [Quit: enebo]
skade has quit [Quit: Computer has gone to sleep.]