KevinCorcoran has quit [Ping timeout: 258 seconds]
subbu is now known as subbu|breakfast
e_dub has joined #jruby
yfeldblum has joined #jruby
yfeldblum has quit [Ping timeout: 244 seconds]
ahhMichael has joined #jruby
toady00|away has joined #jruby
toady00 has quit [Ping timeout: 244 seconds]
cprice404 has quit [Remote host closed the connection]
<headius>
g'day!
<headius>
rtyler: I think the presentation went pretty well...I need to reduce the density somewhat though
nirvdrum has joined #jruby
thsig has joined #jruby
nirvdrum_ has joined #jruby
noop has quit [Ping timeout: 246 seconds]
kares has joined #jruby
nirvdrum has quit [Ping timeout: 244 seconds]
nirvdrum_ is now known as nirvdrum
<nirvdrum>
enebo: That DNS change went through, but the nightly links are still busted.
thsig has quit [Ping timeout: 272 seconds]
thsig has joined #jruby
<enebo>
nirvdrum: yeah we had changed the bucket name some time in the last week which I think neither of us checked while we were travelling
<enebo>
nirvdrum: My last check we get some xml error file saying bucket is wrong name
<nirvdrum>
Any idea when it might work again? I'm trying to debug a 9k w/ rbenv problem with RubyMine and the RubyMine dev can't install it.
<nirvdrum>
Also, master doesn't build.
<enebo>
nirvdrum: well we did not try debugging this more while at JavaOne but how we are saving to s3 is apparently wrong. So we need to figure out why
<nirvdrum>
Okay. Well, if I can be of any assistance, let me know.
<enebo>
nirvdrum: yeah we will get this working but we probably just need some more s3 knowledge
<enebo>
nirvdrum: we are running nightlies and writing it to s3 it is just not working with this url
<nirvdrum>
I've written two different S3 API-compatible implementations, so I have a basic idea of how it works :-)
<nirvdrum>
enebo: Did you set the public read ACL?
<enebo>
connecting with s3hub now
<nirvdrum>
By default, most clients will make the object private so you'd need a signed URL to access it.
<nirvdrum>
subbu|breakfast: It looks like you may have forgotten to commit something. master won't build with your latest changes.
subbu|breakfast is now known as subbu
<subbu>
argh. really?
<subbu>
let me see.
<enebo>
nirvdrum: You know it might just be easier to ask rbenv to change the host
<nirvdrum>
enebo: Could be. It'd have been nice to fix it for anyone with an existing copy. But it hasn't worked for about 3 months, so updating wouldn't be a terrible answer either.
mister_solo has joined #jruby
<nirvdrum>
subbu: [ERROR] /home/nirvdrum/dev/workspaces/jruby/core/src/main/java/org/jruby/ir/passes/AddLocalVarLoadStoreInstructions.java:[115,94] error: incomparable types: Object and boolean
<enebo>
nirvdrum: This works but if I use ci.jruby.org I get NoSuchBucket
<subbu>
interesting ... it builds for me .. java7.
<nirvdrum>
I'm on 1.8.0_20-b26
<enebo>
I also just noticed this says Sep 26 so perhaps our last change is not properly writing to S3 (which was an experiement we forgot to check on)
<subbu>
let me explicitly unbox the boolean.
<headius>
enebo: perhaps it's using domain alias as a root bucket name automatically?
<headius>
the alias record looks fine
<enebo>
headius: I just know it is not working with the alias
tlarevo has joined #jruby
<enebo>
headius: might just be easier to ask rbenv to change it at this point
<headius>
enebo: yeah, I just noticed jruby.org.s3.amazonaws.com. 7519 CNAME s3-directional-w.amazonaws.com.
<subbu>
nirvdrum, i dont have java8 can you do that explicit unboxining on data[0] and verify it fixes it? and, if yes, push the change?
<headius>
so I think it's doing something fancy with the alias
<enebo>
hmmm I might be making a CNAME to a CNAME?
<enebo>
perhaps I should use s3-directional
<enebo>
but the wrong bucketname is weird
colinsurprenant has quit [Quit: colinsurprenant]
<headius>
that's what I mean, the request is probably coming in with domain = ci.jruby.org, which the eventual directional server uses to find root bucket
<enebo>
headius: also you might want to change the name we used back to pre-sept 26
<enebo>
headius: I don’t think the last change is writing to s3 at all now
<enebo>
yeah well it is definitely coming in with ci.jruby.org
<headius>
probably not because I added ci.jruby.org at root and that bucket doesn't exist
<headius>
right?
<enebo>
headius: yeah I think so
<enebo>
headius: but if jruby.org.s3.* is CNAME and also the bucket name it seems like we should be able to do the same thing with another CNAME
<nirvdrum>
subbu: I can do a PR. I don't have push rights.
<enebo>
headius: unless there is some weird registration aspect to s3 I don’t know (which is likely that I would not know)
<enebo>
headius: I can change the CNAME to use s3-directional
<headius>
enebo: Each bucket serves a website namespace (e.g. "www.example.com"). Requests for your host name (e.g. "example.com" or "www.example.com") can be routed to the contents in your bucket. You can also redirect requests to another host name (e.g. redirect "example.com" to "www.example.com").
<Aethenelle>
it's a TODO and has been for a while...
<Aethenelle>
I'm working on a patch...
<headius>
oh, 9112
<headius>
hmmm
<headius>
well it comes down to a question about whether we can do it before they do then
<headius>
we're going to shoot for 2.2 compat so if this gets patched for 2.2 no problem
<enebo>
This is not released software for us and they are fixing it for the same version they are going to put out
<enebo>
heh…
<Aethenelle>
i think that TODO is in a reevaluate state internally with no real drive for resolution... A patch should give them a kick... I missed the last comment on 9112 which adds a bit of a wrinkle
<Aethenelle>
headius: my plan was to create a PR with two commits... one to the new behavior and one back to the old that can be reverted once MRI's behavior changes...
<Aethenelle>
then i thought it'd be easier to just patch MRI...
<enebo>
Aethenelle: seems like 2x the work when you may not need to support one behavior
<enebo>
Aethenelle: If they don’t support it then the second patch is needed
<enebo>
I actually wonder how much this would affect any real world code using prepend? (Is there any real world code using prepend yet?)
JohnBat26 has joined #jruby
<headius>
it could affect all code that uses modules
shellac has joined #jruby
<enebo>
yeah I can see it can affect things but I wonder if it does in practice
<enebo>
lunch…bbiab
<headius>
I'm too scared to risk it :-)
elia has quit [Quit: Computer has gone to sleep.]
<subbu>
headius, when you get a chance, can you run master with JIT and see if everything is still functional?
<headius>
yeah I will do that soon
<headius>
I saw your commits come through
<Aethenelle>
enebo: right, problem is... i'd have to rework most of the PR to match MRI behavior. Either way, some patch is needed...
thsig has quit [Remote host closed the connection]
<Aethenelle>
i also just thought of a problem w/ the code i have... #9112 raises an issue wrt including a module more than once my code assumes once in the hierarchy is all you need... probably should change it to once in the direct hierarchy for includes... not sure the right thing there for prepends...
<Aethenelle>
dfr|work: what it probably should say is that URI's need escaping before they're used as a path element on the file system
<enebo>
I can see how it solves a problem but perhaps if this is his package then he should be using something other than string compare from File.path? like passing it into URI if it is in fact a URI
havenwood has joined #jruby
<enebo>
Aethenelle: I don’t think so
<enebo>
So for non-URI we should match MRI which I believe does no canonicalization
<Aethenelle>
yes, it does... one sec...
subbu is now known as subbu|lunch
<dfr|work>
enebo, hokay, cool. Sounds like you're as confused as I am, so that's good. :)
<enebo>
dfr|work: for URI we could do it but I think it depends on what someone expects
<dfr|work>
enebo, Aethenelle: I obviously want to fix his other issue, but I feel that the stuff mkristian is proposing has too much dependency on URI concepts =/
KevinCor_ has joined #jruby
x1337807x has quit [Max SendQ exceeded]
<enebo>
dfr|work: Personally if non-URI returns exactly what you give it then it seems that it is more consistent to not normalize the path
x1337807x has joined #jruby
<enebo>
Aethenelle: ah file: only too I am betting
<Aethenelle>
enebo: probably any uri... lemme check...
<dfr|work>
enebo, I agree. Also, for URI-related stuff, I'm not sure whether we need to enforce normalization=/
<mkristian>
enebo, yes the space creates a URISyntaxError
<mkristian>
or URISyntaxErrorException
<dfr|work>
mkristian, how about this: 1) Things without uri prefix (file:, etc) do not have a concept of URI at all: things without .jar! are handled as java.io.File only. Things with !/ are handled by however java.util.jar manages to handle them
<enebo>
“A URL is a URI that, in addition to identifying a web resource, specifies the means of acting upon or obtaining the representation”
<dfr|work>
mkristian, and 2) If things do have a URI prefix, then: 1) it has to be absolute (otherwise too much fun with $LOAD_PATH) and 2) Then if opening it as resourcce works
<enebo>
mkristian: yeah makes sense it would give an error
<mkristian>
dfr|work, yes this sounds OK to me.
<Aethenelle>
mkristian: file://he l.rb is not a valid URL either...
toshetm has joined #jruby
<mkristian>
Aethenelle, really ?
x1337807x has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<enebo>
yeah I was just going to type that
<enebo>
space is not valid in URL
<dfr|work>
mkristian, basically, I think it's maybe fine to expose internal implementation behavior, as long as it's reasonable and we try not to fix it too much >.<
<mkristian>
Aethenelle, my knowledge was using java,.net.URL - may that is the difference ;)
statonjr has joined #jruby
<dfr|work>
mkristian, anyhow... how pressing is all of this? :)
<dfr|work>
mkristian, I'm trying to figure out whether I can get to this in couple of weeks, or should allot time for it now =/
<Aethenelle>
URL is a bunch of hacks around URI loosening some rules and making things generally annoying...
<Aethenelle>
mkristian: if the behavior on Java's end for those two is different, I would consider that a bug
<mkristian>
dfr|work, I have no hurry with this - thing actually do work. just my patch for dir glob test with space inside the directory names got a few regex :(
<enebo>
Aethenelle: I believe the weird thing is that URL spec is older than URI from what I recall…
<enebo>
hmm perhaps they were about the same time…things get fuzzy. I sort of worked with someone in those RFCs and I remember some weird conversations
<Aethenelle>
enebo: huh... not only are you right there but It's the other way around... URI is an attempt to make URL sane again...
x1337807x has joined #jruby
<enebo>
In any case, it is all pendantic now.
<Aethenelle>
uri was created as a superset of URL
<Aethenelle>
enebo: but.... pedantry is one of my specialties...
<enebo>
yeah I think this was to represent more than common web tech. Driven by some extent of gopher and hyperg
<dfr|work>
enebo, mkristian, okay, sounds good. I seem to be pretty busy with my 80% these days, so if you guys have something pressing for me to fix, please escalate via direct email and I'll be happy to take a look at it :)
<enebo>
There I think was also maybe something with BEEP as a HTTP replacement as well (god this was a long time ago :) )
<Aethenelle>
2k1 on that... i'm gonna have to read this later...
<enebo>
dfr|work, mkristian: Personally for 1.7.17 I think we really need to make sure we have no more regressions involving load stuff and jar-dependency stuff
<enebo>
This issue seems like it is more of an RFE
<mkristian>
enebo, but jruby-1.7.12 is good enough for the time being
<mkristian>
just newer versions do not behave nice inside OSGi containers
<enebo>
fwiw, I am pretty freaked out about more regressions on 1.7.x now. We are getting a poor reputation at this point. I think some of the refactoring we did was neccesary but it was a difficult pill to swallow. It also showed me that we need many more internal tests for our non-file loading
<enebo>
So I think we need more specs before we can make any significant changes to loading behavior
<mkristian>
enebo, I understand, I can pull it over to 9k - no problem
<enebo>
mkristian: I think that is ok. Hopefully as we add more specs on 1.7 branch we can merge over to master and hopefully this newer code will be working as well
<enebo>
mkristian: At some level we need to make sure people can upgrade from 1.7.x to 9000 without massive breakage (at least for loading of resources)
<enebo>
mkristian: but 9000 is a better place to make these changes and dfr|work is probably a great person to help review this once it is squashed (I will also help look over it)
<mkristian>
the current PR does not break things - at least not nothing which has tests. and the whole thing is already quite old - I did it right after 1.7.14 was released
mister_solo has joined #jruby
<headius>
the previous releases didn't break tests either :-)
<enebo>
:)
tbrock_ has joined #jruby
<headius>
tbh we probably shouldn't have even done dfr's fixes on master, but hindsight
<headius>
er
<headius>
I mean shouldn't have done them on 1.7
<headius>
this will be an ongoing challenge as we continue to maintain 1.7 after 9k release
<mkristian>
there is one more thing: this OSGi ScriptingContainer - there are no tests (beside that is can require 'openssl' and over the last version I comment out even more of those little test
<enebo>
headius: I do think it will make long term maintenance easier though
<headius>
enebo: yeah
havenwood has joined #jruby
<enebo>
headius: It was just unfortunate it uncovered so much
<headius>
we'll need to do more backporting than in the past
<enebo>
yeah
<mkristian>
headius, yes, that is right did not break tests. but that is sign that the tests coverage could be better ;)
<mkristian>
so can I deprecate that OSGiScriptingContainer as such ? or who is going to maintain it ?
nipra has joined #jruby
toady00 has quit [Remote host closed the connection]
donV has quit [Ping timeout: 260 seconds]
<headius>
we could ping whoever contributed it
<headius>
if they're still around they may still be using it
subbu|lunch is now known as subbu
<mkristian>
headius, probably I am asking since the PR we talked about unify the classloading betwenn j2ee, osgi and regular java application. the tests run over 5 osgi right now and the old OSGiScriptingContainer test runs just with felix-4.4
<dfr|work>
headius, I agree about the loadservice changes, I guess =/
KevinCorcoran has quit [Remote host closed the connection]
<dfr|work>
in any case, if we're planning to look into 9k some time in Q1 to be released, I'm absolutely happy to do most of the refactorings there.
<dfr|work>
since that's what matters anyway
<dfr|work>
and if there're any performance adjustements to 1.7, then port them back.
subbu has quit [Ping timeout: 272 seconds]
<enebo>
dfr|work: we just need to probably do smaller refactorings and test the crap out of each change
lanceball is now known as lance|afk
<enebo>
dfr|work: Also do perf work since those refactorings did slow startup time down by like 40%
ephemerian has joined #jruby
<mkristian>
enebo, please you guys do test jruby-complete and the jruby artifact as well. there were so many issues after the first loadService refactoring landed.
mje113__ has joined #jruby
KevinCorcoran has joined #jruby
<mkristian>
and add integration-tests for them as well.
etehtsea has quit [Ping timeout: 260 seconds]
<enebo>
mkristian: yeah I guess we need more automated tests for the resulting artifacts. I manually run something against jruby-complete but it clearly is not enough
<mkristian>
enebo, maybe move some of them to nightly builds since they usually do a lot of downloading ;)
<mkristian>
and not over burden travis so much
<enebo>
mkristian: yeah I think really those level of tests just need to be run occasionally to make sure we do not get surprised after a release
subbu has joined #jruby
<headius>
we have a free hand to set up other runs in jenkins
x1337807x has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<enebo>
We could make some external project which tests artifacts like jruby-complete-tests
<enebo>
I don’t know how that would get flagged to run on SNAPSHOT changes though since we would want it to run when something was published
<enebo>
We can use nightlies with cloudbees too I guess
<mkristian>
headius, is it cloudbees ? I think I read they have maven repo where you can publish your artifacts - can this be the case ?
<mkristian>
enebo, ^^^
<enebo>
mkristian: I think they do but we are publishing to s3 (just still working the kinks out on that though)
<enebo>
mkristian: but we could publish there perhaps and detect changes?
<enebo>
mkristian: Does Jenkins support something like that?
<mkristian>
enebo, but not as maven repo :( so I can not use them from a maven project or gradle project etc
statonjr has quit [Quit: statonjr]
<enebo>
yeah
<enebo>
mkristian: I am unsure
<mkristian>
enebo, I tried to read on how to set them up - but I could not figure if they are public or just private
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] jrubyci pushed 10 new commits to non-indy-jit: http://git.io/HAbwYA
<JRubyGithub>
jruby/non-indy-jit 27f7e05 Charles Oliver Nutter: Fix signatures of instance/class super helpers.
<JRubyGithub>
jruby/non-indy-jit 7b8224c Charles Oliver Nutter: Document why we use real class for ivar table.
<JRubyGithub>
jruby/non-indy-jit 64156fa Charles Oliver Nutter: Add doco for stack layout of super operations.
<enebo>
mkristian: sure although headius has been mucking with config of cloudbees on our jobs
<mkristian>
enebo, headius, if yes, I would cherry pick those build commits from this branch to jruby-1_7 so we can add them to cloudbees (somehow)
subbu has quit [Quit: Ex-Chat]
statonjr has joined #jruby
subbu_ has joined #jruby
subbu_ is now known as subbu
subbu is now known as Guest17
Guest17 is now known as subbu
erikhatcher has joined #jruby
<dfr|work>
headius, completely unrelated. Any chance you guys still have any "JRuby contributor" stickers? I had just obtained a new X1 carbon in lie of my dead one :(
<kares>
mkristian: what would be the issues I can expect to run into (since 1.7.12) on OSGi containers with latest 1.7.x ?
<kares>
.. just using a org.jruby.embed.ScriptingContainer
imperator has joined #jruby
mkristian has quit [Ping timeout: 245 seconds]
tlarevo has quit [Remote host closed the connection]
rcvalle has quit [Ping timeout: 244 seconds]
rcvalle has joined #jruby
toshe_tm has joined #jruby
erikhatcher has quit [Quit: erikhatcher]
toshetm has quit [Ping timeout: 250 seconds]
mkristian has joined #jruby
tbrock_ has quit [Quit: Computer has gone to sleep.]
<mkristian>
kares not sure what went through so once again: ffi native does not load anymore. with the latest version krypt does not load ffi per default anymore. classpath: uri might not work anymore as expected. that is for the OSGiScriptingContainer. the regular ScriptingContainer I have no idea since there are no tests at all inside OSGi. the IsolatedScripting actually works quite OK for felix-3.2/4.4 and equinox-3.6/3.7
<mkristian>
kares, default gems are not recognised as gems with OSGiScriptingContainer or ScriptingContainer but work with IsolatedScriptingContainer
ludyte has joined #jruby
<mkristian>
kares, knoplerfish OSGi framework can not instantiate IsolatedScriptingContainer - some loadService issue - not tested it with other ScriptingContainer
<kares>
mkristian: thanks a lot!
ludyte has quit [Ping timeout: 246 seconds]
phrinx has quit [Remote host closed the connection]
erikhatcher has joined #jruby
e_dub has joined #jruby
lance|afk is now known as lanceball
yfeldblum has joined #jruby
yfeldblu_ has joined #jruby
yfeldblum has quit [Remote host closed the connection]
marr has joined #jruby
elia has joined #jruby
yfeldblum has joined #jruby
x1337807x has joined #jruby
yfeldblu_ has quit [Ping timeout: 244 seconds]
mkristian has quit [Quit: bye]
<enebo>
nirvdrum: yooohoooo
<enebo>
nirvdrum: I do not have rbenv, but it should be fixed now
<enebo>
nirvdrum: The nightlies have not been running successfully for a few days but you should be able to confirm it is fixed by grabbing last successful build
<headius>
dfr|work: no, we pretty much just have the jruby stickers, but it would be good to get our marketing materials up to scratch again
postmodern has joined #jruby
<dfr|work>
headius, I managed to obtain one from a friend who obtained it from some meetup in Wisconsin :)
<dfr|work>
but not that my laptop is dead, I need new one.
<dfr|work>
headius, if I need to pitch in like $10 or whatever, let me know :)
<headius>
subbu: starting to play with your changes on master...I'm getting an NPE in interpreter at this line: ipc = rescueMap.get(instr.getIPC());
<headius>
rescueMap comes back from the scope as null
<headius>
I have not sorted out what scope yet, but it seems like something's amiss
<headius>
a pass messing up rescues or something
<subbu>
headius, hmm ... on what code? i tested rubyspecs and mri tests with a whole bunch of options and seemed to work.
<headius>
this is when I allow the jit to run normally while starting up rspec
<headius>
but it fails inside interpreter
<subbu>
so, this is on code that failed jit and went back to interp?
<subbu>
or pre-jit?
<headius>
unclear...I don't see a fail in the log
<subbu>
k
<headius>
I'll try to figure out what scope it is
<headius>
hmm...maybe it's not the rescue
<enebo>
headius: I was wondering if it was the unboxing frm Integer to int
<enebo>
THE HIDDEN KILLER
<enebo>
(sm)
<headius>
oh yeah, that would do it
<enebo>
I don’t like implicit unboxnig
<headius>
yeah, hard to represent null in primitive
<headius>
that's probably it
<headius>
so the rescue map knows nothing of some instr and this blows up
<headius>
for whatever reason
<enebo>
headius: print the scope out
<enebo>
subbu: we should record passes ran into scope for debugging like this
<enebo>
subbu: we record the data so why not a list of what made it
<headius>
enebo: yeah, doing that now
<subbu>
enebo, makes sense ... back here in a bit ... deploying code right now.
<enebo>
subbu: good luck! :)
<headius>
an explicit list of passes would help us avoid re-running passes we shouldn't, too
<headius>
like ACP
josh-k has joined #jruby
<enebo>
headius: we already won't
josh-k has quit [Remote host closed the connection]
<enebo>
headius: CompilerPass has detection but it asks it dependent passes if it has been performed
<headius>
enebo: if JIT tries to compile a whole file and runs ACP against everything, then fails, future per-method jit may run it (and all other JIT passes) over again
josh-k has joined #jruby
<enebo>
headius: So passes know whether they have been run but we do not record order
<enebo>
headius: but only if there precondition data is wiped out
<enebo>
their
<headius>
also, I tried putting ACP in interp passes for testing and it definitely ran again in JIT passes because I got double frame/scope pushing
<enebo>
wait what is ACP?
<enebo>
heh ok
<enebo>
add call protocall
<enebo>
ACP might be improperly not understanding whether it has run but if it has run it should be invalidated
<enebo>
ah yeah invalidate is empty with a comment saying it can only be run once
<headius>
ok
<headius>
so there is a mechanism at least
<enebo>
actually this code is weird
<enebo>
perhaps I forgot how this is implemented
<Aethenelle>
one more release and I'll have gotten the 2.1 windows cross compile sorted...
<Aethenelle>
no yank this time...
<headius>
yat
<enebo>
I should be possible to call invalidate on a particular pass with the scope and it should wipe out the shit it made from that scope
<headius>
yay
x1337807x has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<enebo>
This design had 2 goals one of which was not to run the same pass twice in a complicate pass list and to make managing the data created by passes simpler
KevinCorcoran has quit [Remote host closed the connection]
<headius>
Aethenelle, enebo: nobu brought up a really good point on 9112: many modules make hierarchy tweaks on include...if module-in-module changes were picked up by the target hierarchy, those tweaks might not be there
<headius>
because they'd only get applied to the module's hierachy
<headius>
since we wouldn't re-include...we'd just assume they're there
mistergibson has quit [Quit: Quitting ... be good to each other :)]
<Aethenelle>
headius: yup... thinking of changing that check to just look at the directly included hierarchy
<Aethenelle>
wait....
<Aethenelle>
i think i just fully got that...
<enebo>
Ohhh make turing machine with prepended and included!!!!!
<enebo>
Endless tape of modules with messages!!!! OMGZ
<Aethenelle>
poo... changing the hierarchy like this is a breaking change...
<enebo>
getting mentally tired…I blame godady and s3
<headius>
I elaborated on that in the bug
<Aethenelle>
i'll work on correcting my jruby patch to match current MRI...
<headius>
if it were just a blob of class data inserted in hierarchy I think it would be ok, but the hook handling might make 9112 break a lot of stuff
<headius>
Aethenelle: tests pass but I'd be worried about how this affects including a Java interface to impl it
<headius>
we do a lot of stuff in hooks for that
<headius>
we may not have good enough tests for modules containing many interface modules...it's a rarely used feature that allows batching together several interfaces and implementing them all by including that one module
<headius>
phooey
<headius>
your way is a lot nicer too
<Aethenelle>
headius: right now, it'd be broken to include a module that includes a java interface expecting to impl it
<Aethenelle>
though.... I think i may have a way to work around that...
<Aethenelle>
if on include you went through the hierarchy calling #included, that'd match current behavior, right?
x1337807x has joined #jruby
<headius>
mayyyybe
<headius>
assuming ancestors worked it might be ok
<Aethenelle>
ancestors has good coverage...
josh-k_ has joined #jruby
<headius>
I assume superclass needs special treatment too
<headius>
there's a lot of code that walks hierarchy expecting to see all included modules inline
<Aethenelle>
i don't think there are any superclass tests...
<Aethenelle>
i'll check that tonight
<headius>
I guess superclass doesn't show modules
<Aethenelle>
i know i've touched ancestors, included_modules, and super on the ruby side
<headius>
only real classes
<Aethenelle>
headius: ahh... then it should be fine...
yfeldblum has quit [Remote host closed the connection]
josh-k has quit [Ping timeout: 240 seconds]
bbrowning is now known as bbrowning_away
yfeldblum has joined #jruby
<Aethenelle>
i at least need to make sure the pseudo superclass created by prepending isn't hit...
<Aethenelle>
I also need to split origin from nonIncludedClass... I was misreading MRI's code for that...
<Aethenelle>
nonIncludedClass is just a pointer to the method table in MRI, origin is the original prepening class...
<headius>
ahh right
<headius>
enebo, subbu: I have the scope
<headius>
ERROR: null rescue target in CLOSURE visit_CLOSURE_1[/Users/headius/projects/jruby/lib/ruby/2.1/optparse.rb:1509]
<headius>
org.jruby.exceptions.RaiseException: (TypeError) can't convert nil into String
<headius>
the second line is the exception that triggered rescue logic
<headius>
that is my bug I believe
<subbu>
k. still deploying.
<headius>
I wonder if this is a fail in begin block or something, and doesn't have a handler
<subbu>
another repo for another project .. back in 15
<subbu>
check isUnsafeScope in IRScope.java
imperator has quit [Quit: Valete!]
<headius>
can do
<headius>
hmm
erikhatcher has quit [Quit: erikhatcher]
mister_solo has quit [Ping timeout: 260 seconds]
KevinCorcoran has joined #jruby
thsig has joined #jruby
<headius>
ok, I think there's a scope getting lost
KevinCorcoran has quit [Ping timeout: 260 seconds]
<nirvdrum>
enebo: The links work at the very least.
<headius>
that's a start!
x1337807x has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<headius>
subbu: ok, I think this is a problem with ACP
<headius>
for a method like this, I'm not seeing a scope creation happening in the outer method body: def foo(id, *args, &block); 1.times { send id, *args, &block }; end
marr has quit [Ping timeout: 272 seconds]
<headius>
it works at command line but probably only because the block and the foo end up sharing the top-level eval scope
drbobbeaty has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<headius>
well, two problems
<subbu>
headius, that is odd .. so, a dyn-scope is being eliminated for foo?
<headius>
subbu: at the very least I'm not seeing pushBinding instructions
<headius>
nor cleanup
<headius>
hmmm
<headius>
actually I'm not sure I even see ACP running
thsig_ has joined #jruby
<headius>
subbu: maybe ACP isn't cascading to nested scopes?
<subbu>
we dont run ACP on blocks right now iirc.
thsig has quit [Ping timeout: 260 seconds]
<subbu>
but, if it is not running on method, it is a bug.
<headius>
yeah something's screwy... I even put a string eval in the body of the method and there's still no push of binding or frame
x1337807x has joined #jruby
<headius>
subbu: I have to call it a day but I will revisit this tomorrow
<headius>
I know these protocols were being added before so something regressed somewhere
<subbu>
k. later.
subbu has quit [Ping timeout: 244 seconds]
subbu has joined #jruby
yfeldblu_ has joined #jruby
yfeldbl__ has joined #jruby
yfeldblu_ has quit [Read error: Connection reset by peer]
havenwood has quit [Remote host closed the connection]
Hobogrammer has joined #jruby
kares has quit [Ping timeout: 244 seconds]
havenwood has joined #jruby
yfeldblum has quit [Ping timeout: 272 seconds]
nipra has quit [Read error: Connection reset by peer]
KevinCorcoran has joined #jruby
KevinCorcoran has quit [Read error: Connection reset by peer]
johnsonch is now known as johnsonch_afk
KevinCorcoran has joined #jruby
statonjr has quit [Quit: statonjr]
yfeldbl__ has quit [Remote host closed the connection]
yfeldblum has joined #jruby
havenwood has quit [Ping timeout: 264 seconds]
nipra has joined #jruby
postmodern has quit [Quit: Leaving]
zeroecco has joined #jruby
Aethenelle has quit [Quit: Aethenelle]
lanceball is now known as lance|afk
havenwood has joined #jruby
colinsurprenant has quit [Quit: colinsurprenant]
brettporter has joined #jruby
josh-k_ has quit [Read error: Connection reset by peer]
yfeldblum has quit [Ping timeout: 244 seconds]
yfeldblum has joined #jruby
tcrawley is now known as tcrawley-away
JohnBat26 has quit [Ping timeout: 272 seconds]
josh-k has joined #jruby
elia has quit [Quit: Computer has gone to sleep.]
rimenes has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
x1337807x has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
thsig_ has quit [Remote host closed the connection]
KevinCorcoran has left #jruby [#jruby]
kotk has joined #jruby
kotk_ has quit [Ping timeout: 272 seconds]
DomKM has joined #jruby
e_dub has quit [Quit: ZZZzzz…]
nirvdrum has quit [Ping timeout: 245 seconds]
x1337807x has joined #jruby
shellac has joined #jruby
ephemerian has quit [Quit: Leaving.]
benlovell has joined #jruby
imperator has joined #jruby
mje113__ has quit [Quit: Connection closed for inactivity]
enebo has quit [Quit: enebo]
colinsurprenant has joined #jruby
x1337807x has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
tbrock_ has joined #jruby
shellac has quit [Quit: Computer has gone to sleep.]
thsig has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] mkristian created detect-root-classloader (+1 new commit): http://git.io/54mfVw
JRubyGithub has left #jruby [#jruby]
<JRubyGithub>
jruby/detect-root-classloader afeea6d Christian Meier: detect the right root classloader for more cases
bbyler_tho has joined #jruby
thsig has quit [Ping timeout: 260 seconds]
thsig has joined #jruby
tbrock_ has quit [Quit: Computer has gone to sleep.]
postmodern has joined #jruby
tbrock_ has joined #jruby
tbrock_ has quit [Client Quit]
baroquebobcat has quit [Quit: baroquebobcat]
josh-k has quit [Remote host closed the connection]
mister_solo has joined #jruby
subbu has quit [Ping timeout: 245 seconds]
x1337807x has joined #jruby
benlovell has quit [Ping timeout: 245 seconds]
mister_solo has quit [Ping timeout: 272 seconds]
benlovell has joined #jruby
x1337807x has quit [Client Quit]
zeroecco has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]