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

Using timezone: Central European Time
<bonniot>frank, how is it going?00:14
<arjanb>is the last testcase in typing\instanceof.testsuite marked correctly?00:24
<bonniot>as PASS you mean?00:25
<arjanb>yes
<bonniot>i think so, why not?00:26
<arjanb>i don't see how instanceof on a type variable makes sense with erasure
<bonniot>right, it's pretty hard to compile00:28
not completely impossible
i think it was bryn who added this testcase00:30
<arjanb>indeed
i think this should not be allowed at least in Nice before 1.0 but the error message can use some improvement00:32
<bonniot>what is it atm?00:33
<arjanb>Class U is not declared00:34
<bonniot>right, we should say something like "type variables cannot be used with instanceof"00:35
<arjanb>i remember you said something about splitting up nice.lang to speed up the testsuite00:43
did you have any succes with that experiment?
<bonniot>I have it locally, it's working OK, but there are still some errors to fix00:44
one side positive point is that it encourages to write more self-contained testcases, that do not rely on nice.lang
<arjanb>where are type vars put in to look up?00:50
<bonniot>in a TypeScope00:52
why?
<arjanb>for improving that error message
i have forgotten quite a few things of nicec's internals00:53
<bonniot>you can do a lookupType on the name00:58
if it returns a MonotypeVar, it's a type variable00:59
<arjanb>i see01:00
that testcase didn't even parse01:11
<bonniot>missing toplevel?01:12
<arjanb>yes
<bonniot>one should always try the testcase before adding the bug marker...01:13
<CIA-3>03arjanb * 10Nice/ (2 files in 2 dirs): Improved error message for typeExpressions.01:31
<bonniot>wow, you're very productive tonight ;-)01:35
<arjanb>not really01:36
<bonniot>?
<arjanb>all small changes
but i'm getting my old pace back :-)01:37
<bonniot>right, but those are also important for the quality of the compiler
yes ;-)
<arjanb>good night01:54
<bonniot>good night
* arjanb leaves
* bonniot leaves02:40
* arjanb joins10:03
* bonniot joins10:49
hi10:51
<arjanb>hi11:01
i'm following the development of antlr 3.0 to see if it's possible to replace the current parser with one generating Nice code11:38
<bonniot>interesting11:39
so how does it look?
<arjanb>nothing is released yet11:41
<bonniot>no beta?11:42
is it in the design phase?
<arjanb>yes
most issues i had about antlr are being improved and adding a backend seems to become easier than with previous versions11:43
<bonniot>cool11:45
<arjanb>http://www.antlr.org/blog/index.tml
<bonniot>what do you expect would be the advantages of switching to antlr 3.0?11:51
<arjanb>possibility of a Nice backend, less limitations in the parser because a new algorithm, making error recovery doable in the parser11:55
it has a low priority and antlr 3.0 won't be ready before the end of the summer i think11:59
<bonniot>do i understand it right that they want to discover the lookahead automatically?12:05
<arjanb>yes12:06
<bonniot>cool!12:07
<arjanb>it's LL(k) for all k at once12:08
<bonniot>are you reading their mailing list?12:17
<arjanb>yes12:18
<bonniot>good12:20
would it autoamtically generate the syntax classes?12:24
<arjanb>no12:25
<bonniot>so there are semantic actions like in javacc?12:30
<arjanb>yes or you can use the default tree generating things12:31
<bonniot>i think that's what I meant12:35
<arjanb>there is something called heterogenous ASTs: http://www.antlr.org/doc/trees.html#Heterogeneous ASTs12:38
but i never tried that12:39
maybe is it possible to create a parser without any sematic action at all13:26
Statement TryStatement() :13:30
{
!"try" Block()
( aCatch() )*
[ !"finally" Block() ]
}
and making the parser so that it creates the standard semantic actions13:31
<bonniot>yes, if you allow customization when necessary13:33
<arjanb>Statement TryStatement() :13:35
{ Statement tryBlock; List<Catch> catches = new ArrayList(); Statement finallyBlock = null;}
{
!"try" tryBlock=Block()
( {Catch c;} c=aCatch() {catches.add(c);} )*
[ !"finally" finallyBlock=Block() ]
{ return createTryStatement(tryBlock, catches, finallyBlock);
}
this looks like a straightforward transformation to me13:36
<bonniot>you could propose this on the antlr lists. good way to know if it already exists, if there is a hidden difficulty, ...13:38
<arjanb>hmm i have already more ideas than i can implement this vacation and the vacation hasn't even started13:54
<fcb>hello13:56
<arjanb>hi13:57
<fcb>you seem to have been busy13:58
any news with the problems in the cvs yesterday?
<arjanb>that has been fixed13:59
<fcb>good work!
<bonniot>hello frank14:00
<fcb>hi
<bonniot>you were not really on the chat yesterday?14:01
<fcb>I was before you were on, but then I went to bed and forgot to disconnect14:02
hmm, I'm still having issues14:04
<arjanb>using the newest nice.jar?14:05
<fcb>ha, actually I'm only having issues with my non-committed nicedoc. lol
<bonniot>what's wrong?14:06
<fcb>in general, or specifically? it won't compile right now because I'm halfway through something.14:07
<bonniot>ok, so that's normal14:08
and what was the difficulty you encountered with doc comments?14:09
<fcb>oh, I need to parse the comment. my first attempt at this was pretty poor. I wanted it to be really flexible, but I made it too flexible such that it thought email addresses where actually tags (eg, @author frank@email.com)14:10
I'm fixing it now14:11
<arjanb>how confusing "Previously called Java 2 Standard Edition (J2SE) 1.5, it's now going by the name J2SE 5.0"14:27
<fcb>that's marketing for you14:30
<bonniot>for whom?14:32
<fcb>for no one in particular. "That's x for you" is an idiomatic expression that means "isn't that typical of x"14:34
<bonniot>ok, i see14:38
didn't know this expression ;-)
<fcb>I don't know ANY French or Dutch expressions!!! I think you are don't pretty well
don't -> doing14:39
<bonniot>wondered if that was another unknwo exp ;-)14:40
<fcb>lol
I'm going to play soccer now - see you later14:42
<bonniot>have a good game!
i need to reboot...15:09
* bonniot leaves15:10
* bonniot joins16:20
<arjanb>hmm what shall i do today?17:40
<bonniot>did you find yet?18:54
<arjanb>no19:01
<bonniot>looking for a small task?19:04
<arjanb>yup
<bonniot>what about handling ; on windows? that's something you can test more easily than me19:07
there could even be a mini-testsuite for windows, testing the different scripts
<arjanb>i find this not an issue19:10
<bonniot>the ; thing on windows?20:37
for the user point of view, it looks like a bug, and was reported as such (by email)20:38
<arjanb>sigh /me dislikes operating systems20:48
<bonniot>good night22:45
* bonniot leaves

Generated by Sualtam