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

Using timezone: Greenwich Mean Time
* GummiEnte joins09:56
<GummiEnte>The invokespecial|onvirtual! has raised some generall discussion: http://gcc.gnu.org/ml/java/2003-10/
<arjanb>maybe they will make it an compile option for gcj10:00
<GummiEnte>Yes, I also think so...
The java version of the logger seems to be more stable than the native compiled version...10:02
<arjanb>any idea why?10:13
<GummiEnte>No. But I haven't had a look at it.10:23
Well, the anon sf cvs-servers now seem to be real faster, but there are no updates from the master... :(10:36
<arjanb>it should not be long until everythings is update if the new servers are in place10:44
* bonniot joins10:46
<bonniot>i've looked at the behaviour of super. it is a bug10:47
caused by the replacement of @ pattern with _ when the tc is the same
this is only an optimization, so we could disable it for now10:48
it's not an optimization in case of primitive types10:49
void foo(int x);
<arjanb>foo(x@int) {}
<bonniot>so either we disable the non-primitive part, or we find another way to do it all10:50
that fails without this code?
<GummiEnte>Daniel, the gcj developer have fallen into a little discussion about the invokeonvirtual issue: http://gcc.gnu.org/ml/java/2003-10/10:52
So, upto the point, when they have enabled both possiblities (as mentioned int eh posts), I'm happy, that you'Äve fixed the invokeonvirtual in the bytecode.10:53
<bonniot>does that concern us? the thing is fixed anyway, no?
<GummiEnte>Yes, it's fixd now.10:54
<bonniot>but they also have made a change?
<GummiEnte>No, not until now, but they may going to do so.
<arjanb>yesterday i wrote a few lines nice code and discovered 2 bugs i think10:55
<GummiEnte>The JVM version of the logger seems not to die, so it must be something in the implementation of the native java.io.Socket that cause the native logger to die... :(10:56
<arjanb>var int[] arr = new int[10];
void main(String[] args)
<GummiEnte>@arjan: a few lines only and 2 bugs... ?!
<arjanb>D:\Nice\.\testbug\test.nice: line 7, column 3:
get(arr, 0) cannot be incremented
yes :(
<bonniot>GummiEnte: so maybe you should report a bug to gcj or classpath about the sockets10:57
<GummiEnte>Hey, you are a windows developer ?!
var int[] count = new int[10];
void main(String[] args)
int x = count[0];
<bonniot>Classpath is the GNU project that implements the java core classes
<arjanb>Ambiguity for symbol count. Possibilities are :
nice.lang.int[] count
<Any T> nice.lang.int count(nice.lang.Collection<T> coll, (T)->boolean test)
<GummiEnte>Ah, I see.
<bonniot>gcj uses mostly their code now
did you try replacing arr[0]++ by arr[0] = arr[0] + 1 ?10:59
<arjanb> arr[0] += 1;
that does work11:00
<bonniot>that's the same, yes
it would be nice to have ++ work too, sure
about count: this is the expected behaviour11:01
<arjanb>i think global var / constant should get priority over methods
<bonniot>there are two toplevel symbols with that name, and overloading resolution is only for method calls and during assignments11:02
but then you could not call the method at all?
<arjanb>only priority in case of assignment11:03
<bonniot>but you example is not an assignment11:04
in an assignment, the type is already used to resolve the ambiguity11:05
<arjanb>you're right but i'm not happy that this code doesn't work as i exspect11:06
<bonniot>yes. in the long term, a more clever overloading resolution could handle it11:10
for now, you can easily work around it, either by assigning count to a local variable, or by renaming it
<GummiEnte>So, I'm again offline. Here in Germany we have got a public holyday...11:17
<bonniot>what for?11:18
you know it?!11:19
Arjan, where do you reside? Wasn't it near the german border?
<arjanb>Enschede about 10 kilometer from the border11:21
So, you could also go on holiday :)
How far would it be for you Daniel?
<arjanb>no it only means that i should not go shopping to today11:22
<bonniot>maybe 600km11:23
<GummiEnte>That's a bit more than you can achieve per bicycle in 20 minutes.
<GummiEnte>Have the nice developers ever meet?
except by irc of course
<GummiEnte>So, if you read the 1.0 that would be a good point, wounldn't it?
<bonniot>yep :-)
<GummiEnte>Ok, my parents will already be waiting... I'll go now...bye
<bonniot>So, freue einheit!11:26
* GummiEnte leaves
<arjanb>i don't see why replacement of @ with _ does make difference for super11:27
<bonniot>because of the order
(in the testcase, the super implementation appears after the calling method)11:28
<arjanb>are you fixing it or should i?11:47
<bonniot>i'm on it11:48
i sent a dev version that fixes it13:55
i will wait to commit changes to cvs, because it is a bit tricky to enable the optim without causing other problems, and I have to leave now
* GummiEnte joins13:57
Hello again
Is the account agreif@users.sf.net not available any longer?
<bonniot>now, it is greifa@...13:58
<GummiEnte>I've forgotton my wiki pw and wanted to reset it, so I should have sent a mail to that mail-address, but I got an mail-dameon...
Oh, ok... Then that should have changed in the wiki...
<bonniot>you can tell him to do that too :-)
<GummiEnte>Right. :)13:59
<arjanb>daniel: what other problems? should i look at that?14:07
<GummiEnte>I just started a native howto, but until I get my new pw, it will reside in the sandbox: http://nice.sourceforge.net/cgi-bin/twiki/view/Sandbox/NativeHowto14:43
So, I'm again offline... CYA14:44
* GummiEnte leaves14:45
<arjanb>i'm going to sleep soon20:46
good night21:10
* arjanb leaves
* bonniot leaves22:26

Generated by Sualtam