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

* bonniot joins09:47
* arjanb joins10:09
* arjanb leaves10:31
* GummiEnte joins10:57
Hello Daniel.
<bonniot>hi
i was not watching...10:58
<GummiEnte>Isn't it any longer possible to compile the actuall cvs version with jdk-1.3?
I just joined.
At me, it gaves me a NullPointerException during 'make bootstrap'10:59
<bonniot>do you get an error
do you also have 1.4?
<GummiEnte>Yes, that is working.
<bonniot>it is true that I recently upgraded to 1.4 as my default jdk
what is the trace?
<GummiEnte>Should I post here?
<bonniot>the top part11:00
<GummiEnte>Stack trace:
java.lang.NullPointerException
at gnu.expr.CopyArgument.compile(CopyArgument.java:34)
at nice.lang.inline.ArrayLength.compile(ArrayLength.java:38)
at gnu.expr.ApplyExp.compile(ApplyExp.java:180)
at gnu.expr.ApplyExp.compile(ApplyExp.java:104)
at nice.lang.inline.NumOp.compile(NumOp.java:82)
at gnu.expr.ApplyExp.compile(ApplyExp.java:180)
Need more?11:01
<bonniot>and what version of Nice do you use to bootstrap?
<GummiEnte>well, I suppose a quite old one: 1170195 2003-06-24 23:17 nice-bootstrap.jar11:02
<bonniot>you need a more recent one
i am surprised it works with 1.4. are you sure?11:03
<GummiEnte>Is it important, if the bootstrapper is compiled with 1.4/1.3?
I'll just give it another try.
<bonniot>you need 0.9.1+, because it fixes a bug with default values using previous parameters, which is now used in the compiler
no, it should not matter11:04
i just did make java=/usr/lib/j2se/1.3/bin/java javac=/usr/lib/j2se/1.3/bin/javac complete, and it worked
do you often compile CVS?
<GummiEnte>Uuups, you are right. 1.4 also did not work. I'm not sure, what leeds me to that point.11:05
Yes, mostly I use the CVS version only.
And now I need nice jdk-1.3 compilant and the latest development snapshot for the .clone() issue.11:08
Ok, with a 0.9.1 bootstraper it works. :) Thnx11:09
<bonniot>good11:16
<GummiEnte>Well, still something not working the way, I wanted it.11:17
The nice/lang/NiceThread.class is not jdk-1.3 compilant compiled?!11:18
That is what I need:
4invokenonvirtual #15 <Method java/lang/Thread.<init> (Ljava/lang/ThreadGroup;Ljava/lang/Runnable;Ljava/lang/String;)V>
But I get this: 6invokenonvirtual #15 <Method java/lang/Thread.<init> (Ljava/lang/ThreadGroup;Ljava/lang/Runnable;Ljava/lang/String;J)V>11:19
<bonniot>this is a constructor introduced in jdk 1.4, right?
that will indeed depend on the jdk you used when compiling Nice11:20
<GummiEnte>I'm again sorry. It's working. It is simply my ant invoking scripts that use JAVA_HOME instead of my given JAVACMD... 11:22
<bonniot>are you compiling Nice with ant?11:23
or your own project?
<GummiEnte>No, my own project... Nice itself just compiles fine with the Makefile
<bonniot>this means that the released version of Nice should always be compiled with 1.3, or it won't be useable, right?11:24
hum, maybe the calls to constructors are generated again when compiling your project, so only that might count11:25
<GummiEnte>Well, it is useable, but not with gcj to compile native...11:27
Should I test, whether the generated constructor depends on the jdk that compiles the nice.compiler?11:28
<bonniot>i see
i think you already did actually, didn't you?11:29
if you were using your own compiled compiler, and by accident used 1.4 for your project
<GummiEnte>Well, I thought it did not work, but it could always be the issue of my Ant invocation scripts.
So, I will try again.11:30
Well, if I want to compile with jdk-1.3 an nicefile with the jdk-1.4compiled compiler, the compiler throws an error...11:35
So I assume that version mismatch does not work...
So, now the other way round...11:36
Ok, the other way round it compiles, but with the jdk-1.4 specific constructor in NiceThread.class. So the nice-lang-classes seems to be regenerated each time...11:37
<bonniot>what is the error in the first case?11:39
<GummiEnte>I saw that you just comitted the dynamic type inferences ... Great. Now I could use an variable after an instanceof as if it is an variable of the instanceof-type, right?
Just have a moment please...11:40
<bonniot>yes, that's the new feature11:41
* arjanb joins11:42
hi
<GummiEnte>hi11:47
Well, complex scenario. Let me try to explain it.11:52
Using JAVAjdk==1.3 and the follwoing does work: NICEC_JAR=/site.usr/local/share/java/nice-0.9.2-pre0-jdk1.3.jar nicec --sourcepath src --destination targets test311:54
Using jdk1.3 with NICEC_JAR=/site.usr/local/share/java/nice-0.9.2-pre0-jdk1.4.2.jar nicec --sourcepath src --destination targets test3 gives me the following:11:55
nice.lang: parsing
test3: parsing
An exception has occured in the compiler
Please fill-in a bug report at the following webpage:
http://sourceforge.net/tracker/?func=add&group_id=12788&atid=112788
Stack trace:
java.lang.Error
at mlsub.typing.FunType.setId(FunType.java:73)
at bossa.syntax.CallExp.getType(CallExp.java:182)
at bossa.syntax.CallExp.wellTyped(CallExp.java:101)
* GummiEnte leaves
* GummiEnte joins11:57
Ok, that was too much... :/
Now it would really be fine to get the really latest development sources...
<arjanb>do you have a few more lines of the stacktrace?11:58
<GummiEnte>at bossa.syntax.OverloadedSymbolExp.resolveOverloading(OverloadedSymbolExp.java:146)11:59
at bossa.syntax.CallExp.resolveOverloading(CallExp.java:242)
at bossa.syntax.CallExp.computeType(CallExp.java:249)
at bossa.syntax.Expression.getType(Expression.java:155)
at bossa.syntax.fun.typecheck(/site.data/ZippedFiles/WinLinux/Java/nice/Nice/src/bossa/syntax:182)
at bossa.syntax.dispatch.typecheck(/site.data/ZippedFiles/WinLinux/Java/nice/Nice/classes/bossa/syntax)
at bossa.syntax.fun.typecheckArgs(/site.data/ZippedFiles/WinLinux/Java/nice/Nice/src/bossa/syntax:97)
at bossa.syntax.fun.typecheck(/site.data/ZippedFiles/WinLinux/Java/nice/Nice/src/bossa/syntax:179)
at bossa.syntax.dispatch.typecheck(/site.data/ZippedFiles/WinLinux/Java/nice/Nice/classes/bossa/syntax)
<arjanb>ok
<GummiEnte>at bossa.syntax.FormalParameters$OptionalParameter.typecheck(FormalParameters.java:228)
at bossa.syntax.FormalParameters.typecheck(FormalParameters.java:359)
at bossa.syntax.MethodDeclaration.typedResolve(MethodDeclaration.java:158)
at bossa.syntax.AST.typedResolve(AST.java:120)
Need even more?12:00
<arjanb>no
<GummiEnte>Not that much pending...
Ok.
Arjan, just for the point, that you did'nt get it: that is only for a jdk-1.3 using a jdk-1.4.2 compiled nice compiler...12:01
<bonniot>just checking: are you sure the order does not matter (the compilation can be different the second time, when the packages are already compiled, unless you use -R)12:02
<GummiEnte>I've removed the destination files...12:03
<arjanb>any idea where this exception happen in the source of your project?12:05
<GummiEnte>Is it a matter of the sourcefile? I haven't had a look at the source up to now.12:07
I#ve reduced the source to:12:09
void main(String[] args) {
double d = -2.1 + 1.0;
System.out.println(d);
\
Still the same...12:10
(Last line should have been a } instead of \
<arjanb>did you build the compiler?12:12
<GummiEnte>Yes, both of them.12:15
<arjanb>i could be a broken build but i have no idea what's going wrong12:20
<GummiEnte>Well, I use the jdk-1.3 version, that will work with an jdk-1.3 and up...12:26
Daniel, did you already have got the time to create the cvs-snapshot system?12:30
<bonniot>yes, it is implemented12:41
<GummiEnte>and how to retrieve the latest version?
<bonniot>one new development version is sent every day, provided it passes the testsuite.
it is the development version12:42
<GummiEnte>That's nice.sf.net/nice.jar?
<bonniot>yes
<GummiEnte>and the source?
<bonniot>do you want to try the instanceof improvement?
<GummiEnte>Yes, I really want to.
<bonniot>the source is in CVS (with the delay unfortunately)
i am sending my own copy12:43
compiled with 1.3, if that makes a difference for you
wait ~1 minute12:44
<GummiEnte>Yes, it makes---
Thnx.
Is it uploaded?12:55
Still someone out there?13:32
<bonniot>yes, i think the one minute is over now! (i was eating)13:36
<GummiEnte>Ok :)13:47
* arjanb leaves14:06
* arjanb joins14:50
<bonniot>GummiEnte: is the test successful?15:13
<GummiEnte>Yes, the instanceof is working perfect, but leads to another problem...15:20
Let me give an example:15:21
I've got a class A and a class B which extends A.
Now I've got a variable a of type A.15:22
Say furthermore B has an field of type A.
So, what I have done up to now was:
if (a instanceof B) a=a.fieldOfTypeA; (with some manual casts).15:23
Now this does not work any longer, because also the lhs must be of type B...
...the lhs requires type B.15:24
<arjanb>this case is similar to ?String foo = ... ; if (foo!=null) foo = foo.some_maybe_null_returning_function();15:30
i guess it's not implemented yet
<GummiEnte>Yes, very similar. I hope this will work in future.15:43
Not only the NULL thing but also my example...
<arjanb>the null thing is already implemented15:44
<GummiEnte>:)15:48
<bonniot>yes, i think we can use the same architecture to implement both, and so handle assignments for instanceof too15:50
<GummiEnte>Great.
<bonniot>i just commited now because thisis already an improvement in most cases15:51
<GummiEnte>Yes, it is. But do you now, when you can commit the other part? 16:12
I'm just thinking, if I should use the brandnew version and adjust my classes or wait for your next commit and then adjust...16:13
<bonniot>i cannot say now when it will be done17:20
it might be a few days, but i cannot commit myself...17:21
arjanb: the collection retypings you added are not in gj.jar?18:28
<arjanb>of min and max not and the other are just for using the method not as static ones18:29
<bonniot>i see18:45
GummiEnte: i'm looking at handling assignment after instanceof18:46
<GummiEnte>:)18:47
<bonniot>i sent a new dev version, that implements it19:26
please test it
<GummiEnte>Ok, I'll do in a minute.19:28
It does not work. Even the new feature of the instanceof does not work.19:35
<bonniot>what is your testcase?19:37
what build do you use?19:38
<GummiEnte>The build you just uploaded?!19:40
<bonniot>can you make sure? with the build date19:41
i'll have to leave...
<GummiEnte>Well, I'll again have a look at it. Maybe we will see us tomorrow.19:42
Where to find the exact build date?
<arjanb>nicec --version
<GummiEnte>(build 2003.08.27, 17:21:55 UTC)19:43
So, this is not the most actuall version I think?!19:44
Well, I also have to leave now... CU20:48
quit
* GummiEnte leaves
<bonniot>god night, arjan23:54
<arjanb>i have been trying to fix the parsing of the range operator but no succes23:57
a..b parser correct but thing like 1..10 not
and the last bugreport of isaac isn't a bug23:58
<bonniot>ok, i will close it tomorrow00:01
<arjanb>goodnight
* arjanb leaves
* bonniot leaves00:02

Generated by Sualtam