Log of the #nice channel on irc.freenode.net

Using timezone: Central European Time
* arjanb leaves00:38
* arjanb joins03:27
* khc leaves06:21
* jeremiah leaves07:08
* arjanb leaves08:13
* CIA-3 leaves10:49
* bonniot joins11:37
* CIA-3 joins11:54
* CIA-3 leaves12:26
* CIA-3 joins12:27
* rubber joins12:39
Hi Daniel!
How are you?
<bonniot>hi12:51
i'm well thanks
<Bluelive>wb
<rubber>Good to hear.
<bonniot>hi bl
<rubber>Hello bl.
<bonniot>what about you?
<rubber>I'm fine too.
I've foudn some ugly bugs at the beginning of the month...12:53
Maybe you've got some time to have a look at it?
<bonniot>the NPE?12:55
<rubber>yes
and there was something else...
<bonniot>on the same day?
<rubber>I've had troubles at 01/02/03.04.0412:56
But on 01.04. there we've found workarounds...
..like:
<V | T <: ?V> ?V get() { ...12:57
But on 02.04. we got something like:
Exception in thread "main" java.lang.Error: at PC 203 then clause grows stack with no else clause
or
Exception in thread "main" java.lang.Error: attempting to push dead variable
and there was the thing with the tuples...12:58
<bonniot>for 01.04, this is a know limitation:
you cannot use ?T inside a class with type parameter T12:59
you should declare get outside the class:
* CIA-3 leaves13:00
<bonniot>(is the idea that T cannot be null?)
* CIA-6 joins
<bonniot>i suppose this is a simplification
i see arjan gave another workaround13:01
<rubber>yes, the other workaround works...13:03
<bonniot>so the mst urgent pb are the code generation ones from 02.04, rightM?13:04
<rubber>...but the worst was the NPE...
Yes, the Errors on 02. and on 03. are the not "NICE"... 13:05
<bonniot>ok
i've got some lag due to my coming back, but i'll look at those13:06
<rubber>Ok, great13:07
The libc is backwards compatible, isn't it?13:46
* arjanb joins14:01
hello Daniel
<rubber>hello arjan
<bonniot>hi!14:12
<arjanb>i'm not sure how to reply to the last 2 posts on the forum14:18
<bonniot>about patches?14:19
<arjanb>yes but mainly the one before that14:20
<bonniot>for the patch, it seems good, don't you think?
<arjanb>yes14:21
<bonniot>could you apply it and answer? I'm starting to look at the NPE rubber reported
<arjanb>ok14:22
<CIA-6>03arjanb * 10Nice/stdlib/nice/lang/collections.nice: similarEmptyCollection for HashSet (from Arthur Smyles).14:50
<arjanb>it looks like i don't have the rights to change patch/rfe things on sf14:51
<bonniot>i'll check that
does it work now?14:55
<arjanb>no14:56
<bonniot>ok, it should be working for patches now14:57
and RFE
<arjanb>it works now14:59
<rubber>What are your plany concerning native and JNI?15:01
<bonniot>you mean to bypass a stage through Java classes?15:02
<arjanb>never discussed that before i think
<rubber>yes...
<bonniot>i think it will happen, but i haven't seen this as a priority so far
what do you think?15:03
<rubber>well, actually the number of "dummy"classes arises... So this was just a question, "if"/"when" native in nice would be available...
<bonniot>it rises because of JNI?15:05
<rubber>yes... I replaced quite a few java functions with c-code though jni; e.g. compress package, crypto package, even some collection parts.15:07
now going to move xml, sockets, ... what will be left is the gui in nice-swt.15:08
I did not believe, that JNI would be that much faster, as it actually is. Ok, I compiled all to native code, but I dropped that, since there have been too many incompabilities...15:09
...(before making the tests with JNI I compiled all to native code...)
<bonniot>what kind of imcompatibilities?15:10
<rubber>gcj on windows with exe/dll boundaries, GC, ... the gcj is not very "stable" on mingw32 target.15:14
at last I didn't even get a running executable :(15:15
for small applications it might be sufficient...15:16
gcj on mingw32 is even worse stable with shared libraries.
<bonniot>did you report the problem?15:17
<rubber>no, not again; its already known.15:18
<bonniot>ok15:19
<rubber>...I could'nt even reduce complexity.
<bonniot>what JDk do you use?15:20
JRE
<rubber>As a static win32 binary, the programs work quite well (but slower than the same java version), but static binaries we are not allowed to distribute...
Aeh, for what?
the application?
several. 1.4.2_0415:21
1.3.1black
<bonniot>wher does the license pb with static come from?15:23
<rubber>has been xerces, gnu-crypto, nice?, poi, swt,...15:24
<arjanb>in classfiles is a native method nothing else then a abstract method with the "native" flag instead of the "abstract" flag
<rubber>I saw, that it must be quite the same, than abstract methods...15:25
...even the method invacation in the bytecode, isn't it?
<arjanb>right
<bonniot>no license pb for nice15:26
<rubber>:)
<bonniot>swt does not allow static redistribution?
<rubber>As far as I remind, no.
<bonniot>When I incorporate a portion of a Program licensed under the CPL into my own proprietary product distributed in object code form, can I use a single license for the full product, in other words, covering the portion of the Program plus my own code?15:38
Yes. The object code for the product may be distributed under a single license as long as it references the CPL portion and complies, for that portion, with the terms of the CPL.
http://www-106.ibm.com/developerworks/library/os-cplfaq.html
<rubber>Interesting.15:41
And, how it goes with the NPE?16:08
<bonniot>i've got an idea what's wrong16:10
nailing it down...
<rubber>:)
Ok, I'll have to leave for today...16:46
bye
* rubber leaves
<arjanb>daniel: reference to a paper of you in last post here: http://forum.java.sun.com/thread.jsp?forum=316&thread=509432&tstart=0&trange=1516:55
<bonniot>you could add a comment saying that it is not just an academic view, since this is implemented in Nice ;-)17:03
<CIA-6>03bonniot * 10Nice/src/gnu/expr/LambdaExp.java: Typo.17:10
03bonniot * 10Nice/ (3 files in 3 dirs): Properly walk over initial values of class fields.18:33
<bonniot>'night23:52
* bonniot leaves23:54
* arjanb leaves00:22