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

Using timezone: Central European Time
<bonniot>arjanb: i was puzzled by the automated build failing, while it all works here01:45
but I think I guessed: the builder did not run once between my commits and yours01:46
so it's using an unfixed version to bootstrap, and that fails like before
funny ;-)
* bonniot leaves03:36
* arjanb leaves05:12
* zzorn joins07:03
* bonniot joins10:52
* arjanb joins14:37
* bonniot leaves16:07
* bonniot joins16:13
* Vurlix joins20:02
<arjanb>hi20:07
<Vurlix>hmm. hi20:23
is nice always going to produce just jvm programs?20:25
since being tied to the jvm kinda limits its usefulness20:26
i guess theres the java-to-native converter...
bleh20:27
* Vurlix leaves21:18
* Artem joins22:56
<bonniot>hi Artem, how are you doing?
could you confirm if my last changes fixed the one testcase you saw failing?22:58
<Artem>hello. i'm pretty busy at work now, ran out of time.23:01
all right, will try to run testcase with these latest changes now.
is it possible to derive singe-argument type from two-argument?23:02
import org.apache.commons.collections.*;
<T,V> Iterator<T> mapIterator(IterableMap<T,V>) = native Iterator IterableMap.mapIterator();
produce23:03
Arguments (template.Hash<java.lang.CharSequence, java.lang.CharSequence>) do not fit:
<T, V> nice.lang.Iterator<T> mapIterator(org.apache.commons.collections.IterableMap<T, V>)
Should probably be23:05
<T,V> MapIterator<T> mapIterator(IterableMap<T,V>) = native Iterator IterableMap.mapIterator();
though its result is practically the same error message.
<bonniot>what is the method call expression?23:09
is template.Hash a subclass of org.apache.commons.collections.IterableMap ?23:10
<Artem>Hash extends AbstractHashedMap extends AbstractMap implements IterableMap...23:16
Invocation is
someMethod( ..., Hash<CharSequence,CharSequence> args ){
let it = args.mapIterator(); ... }
Nevermind the error, is is possible in principle?23:17
<arjanb>what's the code that yield this error?23:20
* arjanb leaves23:22
<bonniot>"derive singe-argument type from two-argument" ?23:26
you mean:
class A<T,U> {} class B<T> extends A<T, String> {}
that's possible
<Artem>no, i didn't meant "derived" as "extends". in commons-collections the Map have a MapIterator which can iterate directly over a Map, whithout the need to create a temporary keySet.23:28
* arjanb joins23:29
<Artem>How to teach Nice that Map<K,V> mapIterator() is returning Interator<K>?23:30
<bonniot>your retyping does that
if you get:
Arguments (template.Hash<java.lang.CharSequence, java.lang.CharSequence>) do not fit:
<T, V> nice.lang.Iterator<T> mapIterator(org.apache.commons.collections.IterableMap<T, V>)23:31
the only possible reason is that template.Hash is not seen as a subclass of org.apache.commons.collections.IterableMap
<Artem>ok,23:34
class Hash<K,V> implements Map<K,V>, IterableMap<K,V> = native Hash;
now it works
thanks23:35
* Artem leaves23:37
* ArtemGr joins23:38
* zzorn_afk leaves23:43
<arjanb>that class retyping shouldn't be needed i think23:45
<ArtemGr>which one? : )23:46
<arjanb>class Hash<K,V> ....
<ArtemGr>yes, does work without it..23:49
i had23:50
class Hash<K,V> implements Map<K,V> = native Hash;
first
<bonniot>isn't it enough if you retype the root of the hierarchy?23:52
<ArtemGr>there are retypings in java.nice of stdlib,23:55
my mapIterator typing wasn't correct at first, and so i thought its maybe necessary..
i have java.lang.VerifyError as it seems on every testcase in the testsuite00:06
(there's no my modifications in the compiler, i've removed them for the test)00:07
[nicec] Exception in thread "main" java.lang.VerifyError: (class: bossa/syntax/fun, method: computeType signature:00:08
(Lbossa/syntax/ExpLocalVariable;)V) Bad type in putfield/putstatic
<arjanb>your compiler is too old to bootstrap00:09
<ArtemGr>i see, will fix now00:10
<arjanb>either download a new dev-version or undo my latest change
<ArtemGr>native\constructors.testsuite was failing? it passes now00:32
no regressions so far. have to go.00:37
* ArtemGr leaves

Generated by Sualtam