<ModusPwnens>
searching for "cannot convert instance of class org.jruby.RubySymbol to class" only yields 3 results on google, so I'm guessing this is an uncommon issue...
pgokeeffe has quit [Quit: pgokeeffe]
bjfish2 has quit [Quit: bjfish2]
bjfish2 has joined #jruby
phrinx has quit [Remote host closed the connection]
phrinx has joined #jruby
bjfish2 has quit [Ping timeout: 245 seconds]
raeoks has joined #jruby
phrinx has quit [Ping timeout: 264 seconds]
nirvdrum has quit [Quit: Leaving]
ModusPwnens has quit [Quit: Page closed]
x1337807x has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
pgokeeffe has joined #jruby
aadam21 has quit [Remote host closed the connection]
<headius>
oh, he left
<headius>
he has a simple typo
havenwood has quit [Remote host closed the connection]
pgokeeffe has quit [Quit: pgokeeffe]
imperator has joined #jruby
pgokeeffe has joined #jruby
djbkd has quit [Remote host closed the connection]
sferik_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
slyphon_ has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
raeoks has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
slyphon has joined #jruby
digitalextremist has joined #jruby
djbkd has joined #jruby
djbkd has quit [Remote host closed the connection]
joast has quit [Quit: Leaving.]
GitHub58 has joined #jruby
<GitHub58>
[jcodings] haozhun opened pull request #11: Add common aliases of unicode blocks (master...block) http://git.io/hBSj
GitHub58 has left #jruby [#jruby]
havenwood has joined #jruby
bjfish2 has joined #jruby
joast has joined #jruby
<bjfish2>
ModusPwnens I would guess you would need to do this: request_obj.some_java_object = @java_object
havenwood has quit []
sferik has joined #jruby
sferik has quit [Read error: Connection reset by peer]
sferik has joined #jruby
subbu has quit [Quit: Ex-Chat]
<headius>
man, this b thing is becoming a total cluster
x1337807x has joined #jruby
x1337807x has quit [Max SendQ exceeded]
x1337807x has joined #jruby
x1337807x has quit [Max SendQ exceeded]
x1337807x has joined #jruby
<imperator>
String#b ?
iamjarvo has joined #jruby
imperator has quit [Quit: Leaving]
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] bjfish opened pull request #2725: [Truffle] Adding args to to_enum call in returnsEnumeratorIfNoBlock annotation. (master...truffle_return_enum_args) http://git.io/hRTt
JRubyGithub has left #jruby [#jruby]
bjfish2 has quit [Quit: bjfish2]
zorak8 has joined #jruby
sferik has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
iamjarvo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
x1337807x has quit [Remote host closed the connection]
x1337807x has joined #jruby
pgokeeffe has quit [Quit: pgokeeffe]
pgokeeffe has joined #jruby
dabradley has quit [Ping timeout: 256 seconds]
dabradley has joined #jruby
dinfuehr has joined #jruby
dinfuehr has quit [Remote host closed the connection]
dinfuehr has joined #jruby
dinfuehr has quit [Remote host closed the connection]
dinfuehr has joined #jruby
zorak8 has quit [Ping timeout: 252 seconds]
kares has joined #jruby
pgokeeffe has quit [Quit: pgokeeffe]
dinfuehr has quit [Remote host closed the connection]
anaeem1_ has joined #jruby
anaeem1_ has quit [Remote host closed the connection]
anaeem1 has joined #jruby
anaeem1 has quit [Ping timeout: 246 seconds]
anaeem1 has joined #jruby
anaeem1 has quit [Remote host closed the connection]
anaeem1 has joined #jruby
anaeem1 has quit [Remote host closed the connection]
anaeem1 has joined #jruby
dinfuehr has joined #jruby
anaeem1 has quit [Remote host closed the connection]
anaeem1_ has joined #jruby
anaeem1_ has quit [Remote host closed the connection]
anaeem1 has joined #jruby
anaeem1 has quit [Ping timeout: 265 seconds]
slyphon has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
yfeldblum has quit [Remote host closed the connection]
kares has quit [Ping timeout: 264 seconds]
x1337807x has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
kares has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] mkristian pushed 7 new commits to master: http://git.io/hRzv
<JRubyGithub>
jruby/master d580ce3 Christian Meier: create usual bin archive as well one where jar files are pack200 compressed...
<JRubyGithub>
jruby/master e366b85 Christian Meier: added check of file size of distribtion artifacts...
<JRubyGithub>
jruby/master 8ae788d Christian Meier: [build] generated pom.xml
JRubyGithub has left #jruby [#jruby]
dinfuehr has quit [Remote host closed the connection]
anaeem1 has quit [Read error: Connection reset by peer]
anaeem1_ has joined #jruby
anaeem1_ has quit [Remote host closed the connection]
anaeem1_ has joined #jruby
bbrowning_away is now known as bbrowning
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] kares pushed 10 new commits to jruby-1_7: http://git.io/h09a
<JRubyGithub>
jruby/jruby-1_7 cfa0011 kares: add a factory onto ClassValue since we're be instantiating the same (based on opt)...
<JRubyGithub>
jruby/jruby-1_7 99a4c30 kares: loadPaths list generics - easier to guess for JavaEmbedUtils consumers
<JRubyGithub>
jruby/jruby-1_7 bf85781 kares: move Java.method_missing into native
JRubyGithub has left #jruby [#jruby]
anaeem1_ has quit [Ping timeout: 246 seconds]
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] aplatypus opened issue #2726: how do I install with Jars? Re: "deprecated. use instead jars/installer" message http://git.io/h0Ht
JRubyGithub has left #jruby [#jruby]
sferik has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] kares closed issue #2595: Issue with object form constructor by reflection. http://git.io/NE2r
JRubyGithub has left #jruby [#jruby]
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] mkristian closed issue #2726: how do I install with Jars? Re: "deprecated. use instead jars/installer" message http://git.io/h0Ht
JRubyGithub has left #jruby [#jruby]
dinfuehr has quit [Remote host closed the connection]
pgokeeffe has quit [Quit: pgokeeffe]
kares has quit [Ping timeout: 246 seconds]
tcrawley-away is now known as tcrawley
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] nirvdrum pushed 1 new commit to master: http://git.io/h0AP
<JRubyGithub>
jruby/master 242c527 Kevin Menard: [Truffle] Simplified some code by using returnsEnumeratorIfNoBlock.
pitr-ch has quit [Quit: Computer has gone to sleep.]
phrinx_ has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] eregon pushed 1 new commit to master: http://git.io/hugN
<JRubyGithub>
jruby/master a00ce58 Benoit Daloze: [Truffle] No such thing as a FrozenError.
JRubyGithub has left #jruby [#jruby]
cyrez123 has joined #jruby
phrinx has quit [Ping timeout: 264 seconds]
iamjarvo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] ralphbellofatto opened issue #2730: jruby test test instructions fail on ppc64 system. http://git.io/huoM
JRubyGithub has left #jruby [#jruby]
Obeliks has joined #jruby
vtunka has joined #jruby
<Obeliks>
I'm running SASS with JRuby, but execution of a 300k input file takes around 2 minutes. This seems much too long for me, so I tried to profile the call: https://gist.github.com/anonymous/489d429b04d9d62e85c4 Unfortunately, there doesn't seem to be an single culprit. Am I reading this correctly?
cyrez123 has quit [Ping timeout: 246 seconds]
mitchellhenke has quit [Quit: Computer has gone to sleep.]
erikhatcher has joined #jruby
pietr0 has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] chrisseaton pushed 2 new commits to master: http://git.io/huMp
<JRubyGithub>
jruby/master 7881f7c Chris Seaton: [Truffle] jt run --igv option.
<JRubyGithub>
jruby/master 2bc8f5c Chris Seaton: [Truffle] Allow GRAAL_BIN to be set per branch by an environment variable.
JRubyGithub has left #jruby [#jruby]
robbyoconnor has quit [Read error: Connection reset by peer]
e_dub has quit [Quit: e_dub]
robbyoconnor has joined #jruby
camlow325 has quit [Remote host closed the connection]
camlow325 has joined #jruby
mister_solo has quit [Ping timeout: 244 seconds]
bbrowning is now known as bbrowning_away
<chrisseaton>
headius: around?
e_dub has joined #jruby
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] chrisseaton pushed 1 new commit to truffle-head: http://git.io/hu7H
JRubyGithub has left #jruby [#jruby]
<JRubyGithub>
jruby/truffle-head c00ff32 Chris Seaton: Merge branch 'master' into truffle-head...
calavera has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
camlow325 has quit [Remote host closed the connection]
e_dub has quit [Quit: e_dub]
Hobogrammer has quit [Ping timeout: 265 seconds]
shellac has quit [Quit: Computer has gone to sleep.]
<kares>
Obeliks: that is likely due therubyrhino being "slow" ... there's better options such as Nashhorn on the JVM but not sure if any of those is usable with ExecJS at this point
<chrisseaton>
Obeliks: are you using SASS just at build time, or while the application actually runs?
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] headius created string-quagmire (+3 new commits): http://git.io/hupQ
<JRubyGithub>
jruby/string-quagmire 6120fca Charles Oliver Nutter: Fix error type.
<JRubyGithub>
jruby/string-quagmire 6ee7103 Charles Oliver Nutter: Fixes/ports to get multi-encoding rindex working correctly.
<JRubyGithub>
jruby/string-quagmire fe821de Charles Oliver Nutter: Re-port tr logic from MRI.
JRubyGithub has left #jruby [#jruby]
<kares>
also ExecJS forces interpreter mode even on Rhino ... last time I tried changing it - it simply went into void ;(
<Obeliks>
chrisseaton: for production we do it on build time, but on developer nodes we run the same code on demand... which makes this a serious issue for us
camlow325 has joined #jruby
sferik has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<headius>
kares: I even fixed Rhino to work better in compiled mode, with too-big methods just staying interpreted
<chrisseaton>
Obeliks: is your code open? I'm looking for examples of people using JS and Ruby together and wanting better performance for a project I work on
<kares>
headius: yy I know
<Obeliks>
chrisseaton: actually, we're not using any JS with Ruby
<chrisseaton>
chrisseaton: I thought SASS was processed using JS?
<Obeliks>
kares: what exactly is ExecJS and how does it come into play here?
<Obeliks>
no, it's a ruby gem
<Obeliks>
at least I thought so
<kares>
Obeliks: how's that possible? i expect for you to be using a JS gem with SASS
<kares>
but I'm not familiar with SASS thus might be wrong ...
<kares>
maybe post a Gemfile.lock
<Obeliks>
kares: if you look at the profiling output, there's nothing that would lead me to assume that the call stack is leaving JRuby
<headius>
Obeliks: looks like it's all Ruby, and I think thiat fits what I know about sass
<headius>
nirvdrum: that String#b thing went quagmire on me
<bjfish2>
nirvdrum yes
<headius>
it was masking a whole bunch of failures, giving false positives
<nirvdrum>
headius: I saw your tweets. It sounded like a mess.
<headius>
I'm going to tag off what's failing on master and go underground for a few days to re-port String stuff that needs it
<Obeliks>
headius: thanks, I will do that
shellac has joined #jruby
yfeldblum has joined #jruby
<nirvdrum>
bjfish2: Nevermind. I think I answered what I was going to ask. Sorry for the noise.
<headius>
chrisseaton: sass would be a good case to test in jruby+truffle, if there's enough implemented to run it
<headius>
any of the pure-Ruby parser/codegen libs
<nirvdrum>
headius: Sucky. Sorry I didn't investigate more when it happened. I thought it was just some silly difference between master and 1.7 at the time.
<headius>
yeah no worries
<chrisseaton>
it seems to have no deps so possible - we've passed 60% of core specs now
<headius>
I was right about the test that was failing...it has always failed and only passed because of #b bug...fixing it just busted open the bug piñata
<nirvdrum>
headius: But I guess this was one nice outcome of having someone else looking into String.
<headius>
yes indeed
<headius>
and most of the functionality I'll be porting over will be needed by jruby+truffle
<headius>
rindex and tr match MRI again
<nirvdrum>
Nice. I apologize in advance for the mess in StringSupport. I'm still hoping to converge on something there.
<nirvdrum>
I also wanted to take some time and run the JVM with inline diagnostics to see if the crazy chain of calls that's currently there is really still needed.
<nirvdrum>
I have to believe that at some point the dispatch cost catches up.
djbkd has quit [Remote host closed the connection]
erikhatcher has quit [Quit: erikhatcher]
djbkd has joined #jruby
yfeldblum has joined #jruby
iamjarvo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
aramisbea has quit [Quit: Leaving...]
pgokeeffe has quit [Quit: pgokeeffe]
<nirvdrum>
lopex: I assume you copied this from MRI, but why does getting a string's length also set its code range?
<lopex>
nirvdrum: because it requires the scan anyways
<lopex>
(well for non fixed char length encodings)
<nirvdrum>
I'm just not clear on what computation it's saving.
<lopex>
next time the core range is known
<lopex>
potentially
<nirvdrum>
Granted, this is contrived, but if I just calculate a string's length in a while loop, the code range is constantly packed, unpacked, and set.
<nirvdrum>
Yeah, but that's my point. It just ignores whatever the code range is anyway.
<nirvdrum>
Unless I'm mistaken.
<lopex>
nirvdrum: well, it might be done conditionally yes
<nirvdrum>
The caller evidently needs to know that StringSupport.strLengthFromRubyString doesn't optimize for single-byte encodings itself.
<lopex>
yea I see the problem in *full
<nirvdrum>
But also, one branch sets the code range, the other doesn't.
<nirvdrum>
If you know what you're doing, it's simple enough.
<nirvdrum>
But we missed a case.
<lopex>
hmm, I think it might be broken after my changes at some point
<lopex>
I dont recall that at all
skade has quit [Read error: Connection reset by peer]
skade has joined #jruby
skade has quit [Client Quit]
tcrawley is now known as tcrawley-away
dcheung has quit [Ping timeout: 246 seconds]
<lopex>
nirvdrum: well, it was quite a bit ago too though :)
<nirvdrum>
lopex: Yeah. I'm just curious :-)
<nirvdrum>
Please don't take it as criticism.
<lopex>
nirvdrum: all criticism will be relayed on MRI
<lopex>
*to
<lopex>
:)
mitchellhenke has quit [Quit: Computer has gone to sleep.]
<lopex>
nirvdrum: but yeah, it's possible I've made a mistake or they do rescan till now
<lopex>
nirvdrum: I think caching the length as rubinius does at least deserves some discussion
<lopex>
enebo also likes ropes but I think it's too much complex
x1337807x has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<enebo>
TOPE ROPES
<lopex>
even without memory model
<lopex>
nirvdrum: I'll look at it tommorrow if you want
<nirvdrum>
Caching would be nice. But invalidating that cache sounds like it'd be very error-prone, unless that operation was pushed into ByteList.
<enebo>
nirvdrum: I have long advocated for length cache but not until we are 100%
<enebo>
nirvdrum: MRI did not do it only because of memory considerations
<lopex>
does rubinius cache hashcode as well ?
subbu|busy is now known as subbu
<nirvdrum>
They do.
<lopex>
java does too :)
<nirvdrum>
Java strings are immutable, so that's a tad easier.
<lopex>
not
<lopex>
they arent
<lopex>
they're just referentially transaparent
<lopex>
nirvdrum: they mutate the hashcode
<lopex>
nirvdrum: in a clever way that doesnt require locking
<nirvdrum>
I've been lied to my entire life then.
<lopex>
it just happens that assigning int is atomic on all java platforms
<lopex>
nirvdrum: look into the source :)
<lopex>
so worse case in java is hash recalc more than once
<nirvdrum>
Sure. But once you calculate a string's length, there's no way to alter that length, is there?
<lopex>
in java ?
<lopex>
in ruby sure there is
<lopex>
that would be a plague of cr-like bugs
<lopex>
but might be worth it
<chrisseaton>
lopex: I think atomic int is guaranteed by the standard actually - but long and double are not
<lopex>
chrisseaton: yeah, that might be the case
<chrisseaton>
i wonder how that works with volatile longs...
<lopex>
chrisseaton: java aligns all things so it might be a bit easier
<enebo>
chrisseaton: and does it luckily differ on x64 which breaks in weird ways because it is not guaranteed by JMM but happens to work
<enebo>
breaks no 32 bit that is
<enebo>
no=on
<lopex>
can it be done via bariers or requires full locking ?
<lopex>
jvm can cheat undeneath
<chrisseaton>
nirvdrum: what about giving byte list a version number? and making it the responsibility of the .length call site to cache, based on that version number?
x1337807x has joined #jruby
<lopex>
chrisseaton: they all mutating methods would invalidate that ?
<nirvdrum>
chrisseaton: You'd still have to track when to update the version number. E.g., chomp and chop use the same byte list and just update the tail pointer.
<chrisseaton>
yeah - they would all bump the version
<nirvdrum>
I guess the number of modifying methods in ByteList is much smaller than the ones in String.
<chrisseaton>
you'd just need to modify each setter - there can't be that many on ByteList?
<enebo>
b
<chrisseaton>
and if the version number gets to INTEGER_MAX you'd consider it 'mega-lengthed'
aadam21 has quit []
<lopex>
enebo: remeber the days when mentalguy entered the conversation on those ocasions ?
<chrisseaton>
and we've ruled out storing the length - not caching it up just always updating it - have we?
<enebo>
lopex: heh
<lopex>
chrisseaton: could same apply to hashcode ?
<nirvdrum>
chrisseaton: On String?
<chrisseaton>
yeah on string
<chrisseaton>
or on bytelist
<lopex>
and cr
<lopex>
etc
<nirvdrum>
Not ruled it out. But I think it'd be very error prone.
<nirvdrum>
We have discussed pushing some of this stuff out of String and into ByteList, since it already has all the necessary info.
<nirvdrum>
But then it's semantically tied to Ruby.
<nirvdrum>
I don't think that's problematic per se.
<chrisseaton>
How hot do we think .length is? Are we spending our time in the right way here?
<lopex>
that would require some statistics from real code
<lopex>
afaik torquebox strugled more with hashcode than length
<nirvdrum>
Well, this started as me asking why calculating the length -- what I would think is a r/o operation -- also updated the code range.
drbobbeaty has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<nirvdrum>
But, string length is used by a lot of other methods.
<enebo>
nirvdrum: length is largely only relevant with non 7bit length since ascii it is byte length
<enebo>
nirvdrum: so you need to ask yourself how common those strings are
<lopex>
there could be a motivation picture "ruby, where string length takes at least 10k lines of code"
<enebo>
nirvdrum: but when they are they are really expensive
<nirvdrum>
enebo: Sure. That just gets back to my original issue.
<enebo>
nirvdrum: this is actually one place where ROPES would help :)
<chrisseaton>
enebo: but knowing that requires the CR to be up to date, which is as much work as keeping a length value up to date
<enebo>
chrisseaton: I don’t disagree. I actually think length should be cached and always have
<nirvdrum>
enebo: In particular, the string length methods in StringSupport expect the caller to *not* call them if single byte optimizable.
<nirvdrum>
enebo: Guess who screwed that up.
<enebo>
chrisseaton: but we have never reached 100% on m17n so I find it difficult to change to it
<enebo>
nirvdrum: hey it’s not nice to point fingers
<lopex>
who!
<nirvdrum>
I was going to say me.
<lopex>
ah
<nirvdrum>
But actually, this is chrisseaton's code. That backfired.
* lopex
ducks
* nirvdrum
ducks too.
* enebo
also ducks...bat!
<nirvdrum>
But this does feed back into the String methods being a crazy chain of calls that maybe 3 people know well and I'm not one of them.
<nirvdrum>
So I was asking about potentially simplifying.
<enebo>
nirvdrum: and some of those people no longer work on our project past irc and don’t quite remember anymore :)
<enebo>
nirvdrum: but I think it is partially about certainty of uncertain semantics
<lopex>
I'm always afraid of weird corner cases of mri
<enebo>
nirvdrum: just passing rubyspecs or even mri unit tests have never filled me with confidence (nothing against the efforts obviously)
<chrisseaton>
it's very easy to accidentally pass RubySpecs
<nirvdrum>
Agreed. That's why I'm so concerned trying to tidy things up.
<enebo>
nirvdrum: but since m17n came into being there is a lot bigger body of direct and indirect tests whch will break if we mess something up
<nirvdrum>
I feel like a bull in a china shop.
<chrisseaton>
'doesn't throw an exception' - that's always an easy one!
<enebo>
I guess at some point if we want a cleaner design we have to take a leap
<enebo>
I would still like us to pass everything available involving strings before we wreck the duct tape of our port :)
<nirvdrum>
It also irks me a bit that obscure encodings can have such a profound effect on design and subsequently, performance.
<chrisseaton>
nirvdrum: can we assume (in the Truffle sense) that they're never used?
<lopex>
nirvdrum: compatibility has it's price
<chrisseaton>
a big assumption that you've never used anything else binary, ascii or utf-8?
<chrisseaton>
i bet that would hold on many many applications
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] nirvdrum pushed 2 new commits to master: http://git.io/h2nB
<JRubyGithub>
jruby/master 5eccb58 Kevin Menard: [Truffle] Added a location for common String guards.
<JRubyGithub>
jruby/master a9f34c8 Kevin Menard: [Truffle] Added a single-byte optimizable form of String#length.
<chrisseaton>
nirvdrum: and I suppose we have to update the code range for other purposes as well anyway?
<nirvdrum>
getCodeRange is a simple int field access (we could push this into the shape).
<nirvdrum>
And maxlength is defined on each encoding. Also a quick int lookup.
<nirvdrum>
chrisseaton: When we modify, yeah. But I updated the translator to create strings with the code range from the lexer. So in the "read-only" case (from source files), we don't need to scan for code ranges at all anymore.
<chrisseaton>
lopex: shapes allow instance variables to be the same as java fields - and then the escape analysis allows them to be allocated on the stack - or in registers even
djbkd has joined #jruby
x1337807_ has joined #jruby
<lopex>
chrisseaton: yeah, I imagine that
<lopex>
headius: has this regex code diverged THIS much ?
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] bjfish opened pull request #2732: [Truffle] Adding frozen check to Array#delete. (master...truffle_array_delete_frozen) http://git.io/h2rw
JRubyGithub has left #jruby [#jruby]
x1337807x has quit [Ping timeout: 245 seconds]
<headius>
lopex: some of it is code style that has changed in JRuby over time, making it hard to see what's missing
<headius>
some methods I audit line by line and some I just wipe out and redo
<lopex>
headius: I can see the diffs..
<headius>
there are some places that don't match at all
<headius>
or they've been refactored so many times I can't see any similarity
<lopex>
headius: the cause of many of those might be the attempt to commonalize 1.8/1.9
<lopex>
well I just made up "commonalize"
djbkd has quit [Remote host closed the connection]
mikecmpbll has joined #jruby
<headius>
yeah there's some of that, and splitting arity paths, and optimizing MatchData use
djbkd has joined #jruby
<headius>
lopex: hey get this...while working on rindex and tr last night I could not for the life of me figure out why exactly-ported code was AIOOB
<lopex>
headius: I wonder if there's a way to interpolate the changes in mri to say "these lines are likely to be changed in near future"
<headius>
turns out several MRI methods allow string byte-wrangling to walk off the string into the \0
<lopex>
yes
<lopex>
those are frequent cases
<lopex>
as many of joni issues as well
<headius>
I think it just happens to work because none of the encodings require more than one additional byte to determine it's an invalid mbc
<lopex>
*as well as
<lopex>
yeah
<headius>
it forced me to get lldb and a debug build of MRI because it didn't make any sense
<headius>
the code shouldn't work, I said
<lopex>
they just bump the pointer
<lopex>
ah
<headius>
so I had to actually step through it and see for myself that it was walking into the null
<headius>
amazing
<headius>
I think I have another such case now
djbkd has quit [Remote host closed the connection]
djbkd has joined #jruby
<lopex>
headius: due to \0 they cant share as much as we do too :)
<headius>
yeah I suppose that's true
<headius>
no cheap slicing if you have to null terminate
<lopex>
no split sharing etc
<lopex>
and that's all for strcpy, strlen, printf friendliness
x1337807_ has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<lopex>
headius: but well, java itself isnt sharing strings anymore
<headius>
MRI is really just a scripting language for C programmers :-D
<headius>
yeah, they ditched that
<lopex>
I understand the reasoning for java, but it's still worth doing in ruby
<lopex>
not sure about arrays though
<lopex>
headius: remember the "another thread sweeping cowed arrays" sporadically thought ?
<headius>
yup
donV has joined #jruby
<lopex>
headius: btw the backrefs can all be reused in things like String#scan etc, the whole joni api is designed around that
<headius>
lopex: ok...I'm not really doing a good job of that in this port, so we may have to rework a bit
<headius>
I will, however, be providing backref and non-backref paths for all of these
<headius>
I think we can get to a point where potential backref methods pass a separate carrier object rather than using the frame
<lopex>
headius: do you use getRegion/getEagerRegion there ?
zorak8 has quit [Ping timeout: 272 seconds]
JRubyGithub has joined #jruby
<JRubyGithub>
[jruby] chrisseaton pushed 2 new commits to master: http://git.io/h2Du