--- Log opened Tue Jul 14 00:00:56 2009
01:04 -!- tgrabiec [n=tomekg@akt29.neoplus.adsl.tpnet.pl] has quit [Remote closed the connection]
03:14 < Eduard_Munteanu> Yay, fixed arguments, will cleanup and submit soon :)
05:41 -!- Eduard_Munteanu [n=edi@79.115.194.125] has quit [Read error: 110 (Connection timed out)]
07:20 -!- penberg_home [n=penberg@cs149038.pp.htv.fi] has quit []
08:25 -!- tgrabiec [n=tomekg@akt29.neoplus.adsl.tpnet.pl] has joined #jato
09:04 -!- penberg [n=penberg@xdsl-83-150-86-196.nebulazone.fi] has joined #jato
09:04 -!- mode/#jato [+o penberg] by ChanServ
09:17 <@penberg> tgrabiec: ping
09:17 <@penberg> tgrabiec: I renamed the Jasmin test to "SubroutineTest.j"
09:17 <@penberg> I hope you're okay with that
09:17 <@penberg> because I'm pushing it now ;)
09:23 < vegard> hi ho :)
09:23 <@penberg> vegard: hi!
09:23 <@penberg> what's up?
09:23 <@penberg> vegard: there's a nice bug report from tomek waiting for you
09:23 < vegard> I'm ready to hack >:)
09:23 < vegard> oh, where?
09:23 <@penberg> vegard: email!
09:24 <@penberg> vegard: "[JATO] invokeinterface bug report"
09:24 < vegard> aha
09:24 < vegard> sorry, diddn't get that far yet :)
09:25 <@penberg> :)
09:32 < vegard> this is the miranda methods bug :)
09:33 < vegard>     // In early VM's there was a bug -- the VM didn't walk the interfaces
09:33 < vegard>     // of a class looking for a method, they only walked the superclass
09:33 < vegard>     // chain.  This meant that abstract methods defined only in interfaces
09:33 < vegard>     // were not being found.
09:33 <@penberg> haha ;)
09:34 <@penberg> vegard: so we have the same bug early jvms did?
09:34 < vegard> yep
09:34 <@penberg> oh, the irony! ;)
09:36 <@penberg> vegard: should be straight-forward to fix?
09:37 < vegard> do we search superclass or superinterfaces first? :)
09:39 <@penberg> superclass?
09:39 < vegard> ah, it doesn't really matter
09:39 < vegard> or rather
09:39 < vegard> yes.
09:40 <@penberg> I am reading invokeinterface specification
09:40 <@penberg> and my head is spinning.
09:40 < vegard> hehe
09:40 <@penberg> If C contains a declaration for an instance method with the same name and descriptor as the resolved method, then this is the method to be invoked, and the lookup procedure terminates.
09:41 <@penberg> Otherwise, if C has a superclass, this same lookup procedure is performed recursively using the direct superclass of C; the method to be invoked is the result of the recursive invocation of this lookup procedure.
09:41 <@penberg> so in tomek's case
09:41 <@penberg> first B then A
09:42 <@penberg> hmmh
09:42 <@penberg> vegard: I'm not sure I understand your problem
09:42 < vegard> yeah, I didn't either ;)
09:43 <@penberg> ;)
09:47 < vegard> hm
09:48 < vegard> interface B has Object as superclass
09:48 < vegard> so while the .java file says "interface B extends A", it gets compiled to "interface B implements A"
09:48 < vegard> so we _do_ need to search implemented interfaces as well
10:07 < vegard> hm. checked proposed changes
10:07 < vegard> still says explicitly "direct superclass"
10:11 < vegard> hm. javap says A is the superclass.
10:14 < vegard> it definitely has Object as superclass
10:16 < vegard> "For an interface, the value of the super_class item must always be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Class_info structure representing the class Object."
10:21 < vegard> ok, think I've got it.
10:23 < vegard> ah, you and your -Warnings
10:25 < vegard> http://pastebin.com/m2c7ad145
10:25 < vegard> any ideas?
10:27 < vegard> (that's jasmin that won't compile a test case, btw)
10:34 <@penberg> vegard: I am using jasmin-sable
10:35 < vegard> me too.
10:35 <@penberg> vegard: [PATCH] vm: fix interface method resolution error
10:35 <@penberg> can I have a regression test for that too?
10:35 < vegard> sure
10:35 <@penberg> penberg@penberg-laptop:~/src/jato$ jasmin -version
10:35 <@penberg> Jasmin version: 2.2.5
10:35 < vegard> Jasmin version: 2.3.0
10:36 <@penberg> hmmh
10:36 <@penberg> tgrabiec: ping pong
10:36 <@penberg> vegard: maybe the format changed?
10:45 < vegard> no, I don't think that's it
10:55 < vegard> I think something is wrong with cup
10:56 <@penberg> did you compile it yourself?
10:57 < vegard> no
10:57 < vegard> sudo apt-get install jasmin-sable
10:57 <@penberg> curious
10:57 <@penberg> ubuntu?
10:57 < vegard> yeah
10:58 < vegard> jaunty I think
10:58 <@penberg> I thought you were a fedora guy!
10:58 <@penberg> I am on intrepid
10:59 < vegard> ah, this my x86_32 ubuntu "server"
11:00 < vegard> ah I'll just try to compile it myself
11:04 < vegard> this isn't working.
11:04 < vegard> ah, they give a jar too.. *tries*
11:06 < vegard> ok, that worked.
11:06 < vegard> hopefully I'll only have to do this whenever a .j file changes
11:09 -!- ahuillet_ [n=user@171.34.83-79.rev.gaoland.net] has joined #jato
11:37 < vegard> hm. the jar manifest file is a bit inconsistent.
11:37 < vegard> *specification
11:38 <@penberg> yeah?
11:38 < vegard> it doesn't specify what the relationship between an attribute and a header is
11:38 < vegard> though in fact I am pretty sure they are exactly the same
11:39 < vegard> and then they special-case the grammar for certain attributes/headers
11:42 < tgrabiec> penberg: pong
11:43 -!- ahuillet_ [n=user@171.34.83-79.rev.gaoland.net] has quit ["Leaving"]
11:43 -!- ahuillet_ [n=user@79.83.34.171] has joined #jato
11:44 < tgrabiec> I'm okay with SubroutineTest, actually I was just about to change the name myself
11:48 < tgrabiec> vegard: I'm on gentoo and I installed jasmin from sources and I don't have 'jasmin' command out of the box. I created ~/bin/jasmin script that executes 'java -jar ~/src/jasmin/jasmin.jar $@'
11:49 <@penberg> tgrabiec: great!
11:49 < vegard> ah maybe I'll just do the same
11:49  * ahuillet_ is out of google money, waiting for the next batch :)
11:55 -!- penberg [n=penberg@xdsl-83-150-86-196.nebulazone.fi] has quit ["Lost terminal"]
12:05 -!- penberg [n=penberg@xdsl-83-150-86-196.nebulazone.fi] has joined #jato
12:05 < penberg> hmhm
12:24 -!- ahuillet_ [n=user@79.83.34.171] has quit [Remote closed the connection]
13:24 < penberg> tgrabiec: nice!
13:25 < tgrabiec> penberg: what is nice ?
13:25 < penberg> tgrabiec: patch series ;)
13:25 < penberg> wide insns
13:25 < penberg> nice to see them getting some attention
13:26 < penberg> tgrabiec: did you see vegard fixed the invokeinterface bug you reported?
13:26 < tgrabiec> penberg: yes, I haven't tested it yet though
13:26 < penberg>   JASMIN   regression/jvm/SubroutineTest.class
13:26 < penberg> SubroutineTest.j:31: Warning - Syntax error.
13:26 < penberg> ^
13:26 < penberg> regression/jvm/SubroutineTest.j: Found 1 errors
13:26 < penberg> hmmh?
13:26 < tgrabiec> what's in that line ?
13:27 < penberg>     ret_w 1
13:27 < tgrabiec> what jasmin version do you have ?
13:27 < penberg> ok, I'm dropping that patch
13:29 < tgrabiec> ret_w is since 2.3
13:30 < penberg> aaah
13:31 < penberg> any way to do that in the one I have?
13:31 < penberg> penberg@penberg-laptop:~/src/tierce$ jasmin -version
13:31 < penberg> Jasmin version: 2.2.5
13:31 < vegard> you have to do the same as us :)
13:32 < tgrabiec> penberg: apparently wide ret doesn't work in earlier versions
13:33 < penberg> hmm
13:33 < penberg> well, I guess I need to upgrade then?
13:34 < penberg> although I'd prefer not to do that manually
13:34 < penberg> hmm mmhmmm hmmm
13:35 < tgrabiec> penberg: I looks that 2.3 introduces _w postfix for some instruction, so we will need that version to write wide insn tests
13:36 < tgrabiec> installing jasmin from sources is just downloading them and moving in the right place
13:37 < tgrabiec> and making the 'jasmin' script
13:38 < penberg> hmmhm
13:38 < penberg> can we have a "make jasmin" thing that downloads the sources and sets everything up?
13:39 < penberg> side note: doesn't jasmin allow us to hardcode the bytecode in hex?
13:39 < vegard> pekka... ;-)
13:39 < tgrabiec> does it?
13:40 < penberg> I don't see anything in the manual :(
13:40 < tgrabiec> yeah, me either
13:41 < penberg> +    astore 1
13:41 < penberg> +    ret 1
13:41 < penberg> now we have jsr_w at least
13:41 < penberg> so I'm crippling the test a bit to avoid installing new jasmin
13:41 < penberg> sorry
13:42 < tgrabiec> penberg: :)
13:42 < penberg> tgrabiec: we need to figure out how we're going to distribute jasmin with jato
13:42 < penberg> if we want to bump up to 2.3
13:42 < penberg> OTOH, we can probably just switch over after a year or so
13:42 < penberg> when it's reasonable to expect people to be running ubuntu 9.04
13:42 < penberg> or later
13:43 < vegard> why not just require 2.3 ? we _do_ require it, apparently.
13:43 < penberg> because that doesn't fix the "make works out of the box" problem!
13:44 < penberg> I don't want to set up anything manually because that means _everybody_ needs to do that
13:44 < penberg> and on practical level, I'm not upgrading to 9.04 right now for various good reasons.
13:44 < vegard> well, make never worked out of the box without the right libraries
13:44 < penberg> yes it id
13:44 < penberg>     $ sudo apt-get install ecj classpath libffi-dev binutils-dev libzip-dev jasmin-sable
13:44 < penberg> I don't know why you want to argue on this
13:45 < penberg> just saying we require something fixes _nothing_
13:45 < vegard> well, my 2.3 version from ubuntu doesn't even work
13:45 < vegard> so I need the custom script anyway
13:45 < penberg> well, that's bad, isn't it?
13:45 < penberg> so we need to fix it somehow
13:46 < penberg> make jasmin, probably?
13:46 < penberg> wget <url>
13:46 < penberg> install to $HOME/bin
13:46 < penberg> tgrabiec, vegard: right?
13:47 < tgrabiec> penberg: make jasmin could be nice, but we also have to ensure that ~/bin is in $PATH
13:47 < vegard> not install to $HOME/bin
13:47 < penberg> vegard: where then?
13:47 < vegard> I have all my local programs and libs in $HOME/local/
13:47 < penberg> $JATO/tools/jasmin ?
13:47 < vegard> so it shouldn't be hard coded at the very least
13:47 < penberg> ok ok, agreed.
13:47 < penberg> lets install it in tools/jasmin then
13:47 < vegard> why not just download the jar and use that?
13:48 < penberg> vegard: put that in jato?
13:48 < vegard> i.e. so java -jar jasmin-2.3/jasmin.jar <options>
13:48 < penberg> yes, download jasmin.jar to tools/jasmin/
13:56 < vegard> so for -jar handling, you think it's enough to prepend the jar file to the classpath after figuring out which one is the main class?
13:58 < penberg> vegard: yup
13:58 < penberg> not sure if MANIFEST can refer to other jars too
13:58 < penberg> dunno
15:00 -!- penberg [n=penberg@xdsl-83-150-86-196.nebulazone.fi] has quit ["Lost terminal"]
15:03 -!- penberg [n=penberg@xdsl-83-150-86-196.nebulazone.fi] has joined #jato
15:04 < vegard> heh
15:04 < vegard> wondering if I should make this jar/manifest stuff into its own library
15:04 < vegard> it's completely separate from anything jato-specific
15:08 < penberg> cafebabe?
15:12 < vegard> doesn't really have anything to do with the class file either
15:14 < vegard> ./jato -jar frozenBubble.jar -cp /usr/share/classpath/glibj.zip
15:14 < vegard> __convert_ldc: warning: unknown tag: 7
15:14 < penberg> well, I am not sure what's the benefit of adding even more libs?
15:14 < penberg> yay!
15:18 < vegard> ouch, that's just above 700 lines for the jar/manifest support
15:18 < vegard> I
15:18 < vegard> 'll send the patch, sounds ok?
15:19 -!- tgrabiec [n=tomekg@akt29.neoplus.adsl.tpnet.pl] has quit [Read error: 110 (Connection timed out)]
15:21 < penberg> vegard: yes pls
15:21 < penberg> what the hell happened to github
15:22 < penberg> vegard: is github broken for you too?
15:23 < vegard> don't know
15:26 < vegard> yeah, it doesn't connect
15:26 < vegard> was like this yesterday too iirc
15:26 < vegard> (at some point, I mean, and then it obviously was better in the meantime)
15:26 < penberg> :(
15:26 < vegard> but I sent the patch, going home now
15:27 < vegard> see you tomorrow :)
15:27 < penberg> see ya
17:46 -!- ahuillet_ [n=user@AToulouse-158-1-87-122.w90-11.abo.wanadoo.fr] has joined #jato
18:04 -!- user__ [n=user@AToulouse-158-1-87-122.w90-11.abo.wanadoo.fr] has joined #jato
18:06 -!- ahuillet_ [n=user@AToulouse-158-1-87-122.w90-11.abo.wanadoo.fr] has quit [Read error: 60 (Operation timed out)]
18:20 -!- user__ is now known as ahuillet
18:30 -!- tgrabiec [n=tomekg@bdk197.neoplus.adsl.tpnet.pl] has joined #jato
18:31 < tgrabiec> penberg: so the solution for jasmin 2.3 to work is adding jasmin.jar to tools/jasmin ?
18:32 < tgrabiec> I have 'wide' support ready but my jasmin tests require 2.3 for istore_w etc.
18:37 -!- Eduard_Munteanu [n=edi@79.115.194.125] has joined #jato
18:37 < Eduard_Munteanu> Hi.
18:41 -!- tgrabiec [n=tomekg@bdk197.neoplus.adsl.tpnet.pl] has quit [Remote closed the connection]
18:54 < Eduard_Munteanu> penberg, I got reg parameters working, just a few issues to sort out and clean the patches.
18:54 < Eduard_Munteanu> However, the LIR shows no passing of *this. Is this normal behavior?
18:55 < Eduard_Munteanu> I figure we determine *this automatically, right?
18:56 < Eduard_Munteanu> (I'm asking this because the JVM specs tell about local_index 0 being assigned to *this.)
19:13 < ahuillet> Eduard_Munteanu : the LIR should show it I think
19:16 < Eduard_Munteanu> ahuillet, it seems we don't pass *this at all. Perhaps we can obtain it otherwise?
19:16 -!- ahuillet_ [n=ahuillet@AToulouse-158-1-87-122.w90-11.abo.wanadoo.fr] has joined #jato
19:16 < Eduard_Munteanu> ahuillet_, it seems we don't pass *this at all. Perhaps we can obtain it otherwise?
19:16 < ahuillet> well I'm no expert here
19:17 < Eduard_Munteanu> ahuillet, okay, will leave it at that for now. Will fix later if it's a problem indeed.
20:19 -!- tgrabiec [n=tomekg@bdk197.neoplus.adsl.tpnet.pl] has joined #jato
20:21 < tgrabiec> Eduard_Munteanu, can you show me the trace on which you think we don't pass 'this' pointer ?
20:30 < Eduard_Munteanu> tgrabiec, http://pastebin.com/d7e20389f
20:30 < Eduard_Munteanu> tgrabiec, but ignore the disassembly and the fact we don't read stack results correctly.
20:30 < Eduard_Munteanu> s/results/params/
20:31 < tgrabiec> and can you show the java source for ParameterPassingTest ?
20:32 < Eduard_Munteanu> tgrabiec, here it is: http://pastebin.com/d6d662791
20:33 < Eduard_Munteanu> Even if add() is static, there should be an instance of the class, therefore a *this, right?
20:33 < tgrabiec> ok, so ParameterPassingTest.add method is static, so we don't pass 'this' pointer to it
20:34 < tgrabiec> no, there is no this pointer nor a class object pointer for static methods
20:34 < tgrabiec> there is one for JNI methods
20:34 < tgrabiec> (static JNI methods)
20:34 < Eduard_Munteanu> tgrabiec, oh, I see, thanks.
20:39 < Eduard_Munteanu> tgrabiec, I'm kinda || with java.
20:40 < ahuillet> || == ?
20:40 < Eduard_Munteanu> ahuillet, paralell, having nothing in common :
20:40 < Eduard_Munteanu> :)
20:40 < ahuillet> oh, you too ;)
20:40 < tgrabiec> :)
20:40 < Eduard_Munteanu> ahuillet, yup :)
20:51 < Eduard_Munteanu> Hm, it was not as I expected, correct arg passing will need a bit more work.
20:53 < Eduard_Munteanu> That is, it still doesn't work right when mixing stack and reg params.
20:54 < Eduard_Munteanu> ahuillet, it is painful, like you said.
21:05 -!- ahuillet [n=user@AToulouse-158-1-87-122.w90-11.abo.wanadoo.fr] has quit ["Leaving"]
21:05 -!- ahuillet_ is now known as ahuillet
21:09 < ahuillet> ok, now I remember the passw
21:09 < ahuillet> *pass
21:09 < ahuillet> *password for my bouncer
21:17 -!- tgrabiec [n=tomekg@bdk197.neoplus.adsl.tpnet.pl] has quit [Remote closed the connection]
22:44 -!- penberg_home [n=penberg@cs149038.pp.htv.fi] has joined #jato
23:11 -!- penberg_home [n=penberg@cs149038.pp.htv.fi] has quit []
--- Log closed Wed Jul 15 00:00:56 2009