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

Using timezone: Central European Time
* ArtemGr joins08:34
* bonniot joins10:22
<Bluelive>hey10:29
<bonniot>hi10:32
<Bluelive>attemtion at ibm, nice10:35
you should links to that stuff on the nice.sf page
+put10:36
<bonniot>what stuff?10:37
<Bluelive>http://www-106.ibm.com/developerworks/library/j-alj10064.html10:38
links like that look good on a projects homepage ;)
hmz from october, sorry it come in as new on java.blogs10:39
s/come/came/ ugh im not awake
<bonniot>you're right, it should be linked to10:51
* ArtemGr leaves11:28
* zzorn joins11:37
<bonniot>i'll be back later...11:55
* bonniot leaves
* bonniot joins12:44
* bonniot leaves12:55
* arjanb joins14:06
* bonniot joins14:30
arjanb: has most of the conversion been done during january?14:45
<arjanb>largest part in december14:46
<bonniot>strange because in nicec from 2004.12.29, SDE was only 20641 bytes14:47
what has made it grow until now?14:49
<arjanb>maybe i'm confused14:50
* fcb joins
<arjanb>hi14:52
daniel was that nicec built from an updated cvs version?14:53
<CIA-1>03bonniot * 10Nice/src/gnu/bytecode/SourceMap.java:
Truncate SourceDebugExtension to make sure it does not exceed 65535 bytes,
to workaround a Sun JVM bug.
<bonniot>hard to know now...
<fcb>hello14:55
<bonniot>hi frank
arjanb: I think it is, because for instance it has readImportedAlternative in Nice, which was converted Dec 21
<fcb>daniel: have you tried to reproduce my problem14:56
<bonniot>sorry, not yet
<fcb>ok, no worries
<bonniot>so please give me instructions ;-)14:57
<fcb>oh, ok14:59
these are the steps I use:15:01
1) "make complete" (can probably skip this step) - no problems
2) "make nicedoc" - no problems, but no suprising
(optional step 2a: "nicedoc --sourcepath src/ -d /tmp/niceout nice.tools.doc" has two warnings)
3) "touch src/nice/tools/doc/htmlwriter.nice"
4) "make nicedoc" fails - very surprising for me.
* Bluelive leaves15:05
<bonniot>one moment, i'm on the phone (over IP ;-))15:07
<fcb>np
<bonniot>i'm doing 4) now15:09
The implementation test failed for method nice.lang.void generate(Compilation compilation, String mainPackage, ?File outDirectory):15:11
no alternative matches (bossa.modules.Compilation, java.lang.String, null)
<fcb>yes, that's what i get15:12
<bonniot>ok, so I can reproduce it15:13
either I or arjan will fix it15:14
<fcb>brilliant :)
<bonniot>you can work around by issuing 'make' in between, right?
<fcb>yep
<bonniot>ok
<arjanb>maybe it has to do with overrides in different packages?15:15
<bonniot>do you have other issues to discuss?
arjanb: good idea
<fcb>yes
<bonniot>arjanb: the first step should be to reproduce it in a small testcase
<fcb>is the output of nicedoc and javadoc anywhere on the web (ie, for the entire lang)
<bonniot>fcb: fire on
for the compiler source you mean?15:16
<fcb>yeah
<bonniot>no
<fcb>would it be feasible to do so?15:17
<bonniot>not very practical because SF does not have cron jobs at the moment15:18
is it hard for you to do it locally?
<fcb>no, not hard15:19
just thought it was a good idea
I might as well ask one last question while I'm here15:20
<bonniot>i'm not even sure if there is a working target in the Makefile for it
<fcb>actually, forget my last question, I think I just answered it for myself15:22
thanks, guys
<bonniot>no pb
what are you up to at the moment?
<fcb>same as usually - I want to have many more hyperlinks15:23
<bonniot>good
arjanb: are you trying to reproduce the bug?15:25
<CIA-1>03bonniot * 10Nice/src/bossa/syntax/javaFieldAccess.nice: Cleanup.15:44
<fcb>this is from methodDeclaration.nice15:48
syntacticConstraint != null ? notNull(syntacticConstraint)
: mlsub.typing.Constraint.toString(notNull(this.getType()).getConstraint()))
+ String.valueOf(this.getReturnType()15:49
what is meant by getContraint()? do some methods not have return types?
<bonniot>they all have return types
the constraint is what's inside <...>15:50
like <Foo T>
<fcb>ok
<arjanb>was away for a moment15:58
the generate method exists only in the nice.tools.doc package
* zzorn joins16:04
<bonniot>yes, it's definitively a bug16:11
* zzorn_afk leaves16:17
<arjanb>deleting the compiled things of nice.tools.doc doesn't seem to have any effect16:45
<fcb>ok, I'm off. thanks for looking at my problem16:52
good luck
* fcb leaves
<bonniot>Link test for nice.lang.void generate(Compilation compilation, String mainPackage, ?File outDirectory)16:59
Alternative: generate(bossa.modules.Compilation compilation, java.lang.String mainPackage, java.io.File outDirectory)
Multitag: bossa.modules.Compilation, java.lang.String, null
Multitag: bossa.modules.Compilation, java.lang.String, java.io.File
the problem is that the alternative has a non-null pattern for File
<arjanb>but where does that alternative come from?17:01
<bonniot>it must be the default implementation
i think I start to see17:26
<arjanb>:) i get only more confused
<bonniot>given that nicec thinks there is an override, methodImplementation.addPatterns is called17:27
this adds patterns to the default impl, which all become non-null
<arjanb>so how can it find an override while there's only a single method with that name17:29
<bonniot>that's one problem
but i think there is another one17:30
YES, there it is:17:31
/// PASS
/// package a
/// Toplevel
abstract class A
{
void foo(?String s);
}
/// package b import a
/// Toplevel
class B extends A
{
override void foo(?String s) {}
}
[nicec] The implementation test failed for method nice.lang.void foo(a.A this, ?String s):
[nicec] no alternative matches (b.B, null)
<arjanb>this looks familiar17:32
i think there was a bug like this (maybe in single package case)17:34
<bonniot>that was fixed?17:38
<arjanb>i think so
<bonniot>this one also fails in a single package
<arjanb>unless my memory is failing there was something else causing no alternative matches (..., null ) before17:40
<bonniot>there was what caused the introduction of NotNullPattern17:41
<arjanb>that could be it17:42
viewcvs is down :(
<bonniot>yes
cvs is not though
i think this should work:17:44
if (tc == null || !sure)
return new VariablePattern(name: name, loc: loc);
pattern.nice:266
the question is what happens if you override ?A with ?B17:45
but if i remember we decided to disallow it
<arjanb>indeed
<bonniot>so i guess that's an acceptable fix17:46
yes, it does fix the testcase
nicedoc still has the override warning of course, but it does not fail anymore17:52
<arjanb>that change looks correct to me assuming ?B overriding ?A is disallowed
<bonniot>i committed the change. it looks like cia is on strike...19:00
<arjanb>yeah cia isn't online often last weeks19:13
* CIA-1 leaves20:16
* CIA-temp108 joins20:18
<bonniot>well it was online, just didn't say anything21:30
see you later...21:52
* bonniot leaves21:55