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

Using timezone: Greenwich Mean Time
<Bluelive>hmz02:19
okee ik volg de code niet echt meer02:20
morgen maar eens kijken of een rewrite nodig is
<arjanb>woei het werkt eindelijk
taktiek van copy pasten en random dingen veranderen werkt soms wel02:21
<Bluelive>cp'en probeer ik te vermijden
random dingen veranderen wordt het altijd wel als je moe wordt
maar als hij is goed keurd wat gewoon fout is02:22
tsja
<arjanb>ik kan beter nu eerst maar gaan slapen02:25
* arjanb leaves
<Bluelive>well03:11
constraints seem to work
only collapsing constraints need to be accepted
* arjanb joins11:14
<Bluelive>whats LtU's problem, its allways so slow11:25
<arjanb>why does readObject need to be private :(13:48
<Bluelive>geen idee13:55
als jullie een methd hebben met <R> R next()16:06
hoe bepaal je dan R vaan new R
<arjanb>new R is niet mogelijk16:07
<Bluelive>hmz
bij class A<B> kan je wel new B doen16:08
<arjanb>new R[10] is wel mogelijk in Nice maar niet in java 1.5
bah slome testsuite16:17
<Bluelive>?16:19
wat voor testsuite gebruiken jullie ?
<arjanb>een simpel eigengemaakt ding
<Bluelive>hoe ziet het eruit16:20
<arjanb>maar doet er nu bijna een kwartier over alle 950 tests
gewoon allemaal testcase in een aantal tekst files16:21
<Bluelive>maar waar op check je dan ?
<arjanb>of de compiler niet crashed en of een compile error krijgt en waar16:22
<Bluelive>geen testcases om te testen of iets wel failed ?16:25
<arjanb>wel
class A { int x; }16:26
class B extends A { int /* /// FAIL HERE */ x; }
class C extends B {}
_/// FAIL16:27
_ /// Toplevel
_class A { int x; }
_class B extends A { int /* /// FAIL HERE */ x; }
_class C extends B {}
_/// PASS bug
_ List<List<int>> x = [[1,2,3]];
_ int i = x[0][1];
<Bluelive>ik begin c++ template systeem steeds meer te waarderen16:57
gewoon over z'n simpelheid
<arjanb>implementatie daarvan is niet simpel16:58
<Bluelive>simpeler16:59
hmz17:36
constraints collapsen niet correct naar extends
* bonniot joins19:35
hi
<arjanb>hi19:36
just fixed a rare bug in gnu.bytecode.Type19:37
<bonniot>i did not see any testcase
<arjanb>have made a bugreport for kawa too
<bonniot>i wonder, because in Nice char is not a subtype of int
<arjanb>no19:38
<bonniot>(for kawa, did your check their CVS code? it might have changed since)
<arjanb>i have checked that
<bonniot>ok
so what is the Nice testcase?
<arjanb> assert foo(("abc",('x',3)))==3;19:39
int foo((String s,(char c, int i))) = i;
('x',3) was compiled as an array of Object
<bonniot>ok, so it was not optimal?19:40
<arjanb>and the method expected an array of int because of (char c, int i)19:41
<Bluelive>wow, tupels look totally confusing ;)
<bonniot>(i see you commited the testcases too, good!)
ah ok
that's the standard notation
but you rarely explicitely nest several levels at one point19:42
it's usually:
(int x, int y) = foo(...);
return (x+1, y);
which is I think quite readable :-)
<Bluelive>;)19:43
is int equivalanet to (int, void) ?
or even (int) ?
<arjanb>no
yes but (int) isn't a tuple
we only have nested tuples to make parsing complex :)19:44
<bonniot>:-))
<Bluelive>:)))19:45
<arjanb>daniel see the log of yesterday for problems GummiEnte has with serialization
<bonniot>i quickly looked at the other commit messages
i know, i read them and he wrote an email to me19:46
i saw you implemented te serialUID
<arjanb>yes
<bonniot>that's a good point. there is only read/writeObject to add19:47
(in NiceClass you left a commented out call to createSerialUIDField())
<arjanb>forgot to remove that19:48
<bonniot>try to check the diff before commiting
but it's not a big problem :-)
<arjanb>i do but overlooked the comment this time19:49
<bonniot>no problem
the getters/setters are also good because they allow writing java beans in Nice :-)19:51
<arjanb>the getters/setters are not called yet from Nice code
<bonniot>i know19:52
but i think we should in the future19:54
so for instance we can do:
package a;
class A
{
<Bluelive>well please use a nice properties syntax ;)
<bonniot> int x;19:55
int foo() = x + 1;
}
package b;
class B
{
x() { ...; return x; }
}
what do you propose?19:56
<Bluelive>modifiers type shortidentifier (get function)? (set function)?19:57
with get and/or set its a property19:58
<bonniot>for use, it should be exactly like using fields
for definition, what about
x() { /* getter */ }19:59
x(value) { /* setter */ }
* bonniot is away: send me a private message if needed
<Bluelive>?
why not my syntax
<arjanb>is more verbose20:01
<Bluelive>more ? its less verbose :)
atleast, i think it is20:06
<arjanb>for this the parser doesn't have to be changed20:09
<Bluelive>!@#%#@%#%!@20:20
why wont it wrk
<bonniot>arjan, i'll implement the private read/writeObject21:40
<Bluelive>and it works :)21:41
<arjanb>how?
what about a real interface for Serializable with read/writeObject and read/writeResolve21:42
and then some trick to make read/writeObject private21:43
?21:48
<bonniot>i have implemented the tricks part for read/writeObject22:11
i asked christian to test it
(not in CVS yet, it is not polished)
i'm not sure what advantage an interface would bring
i'm going to sleep on this :-)22:12
good night
* bonniot leaves

Generated by Sualtam