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

Using timezone: Central European Time
* arjanb leaves05:57
* Bluelive leaves08:16
* bonniot joins09:13
* CIA-3 leaves10:47
* arjanb joins11:10
* CIA-3 joins11:18
<arjanb>i can't get nicedoc running11:39
<bonniot>hello. you should contact Frank then12:02
i did not try that version yet
<arjanb>i probably doing something wrong with classpaths13:09
but the error is not helpfull:
Exception in thread "main" java.util.zip.ZipException: The system cannot find the file specifie
at java.util.zip.ZipFile.open(Native Method)
at java.util.zip.ZipFile.<init>(ZipFile.java:112)
at java.util.jar.JarFile.<init>(JarFile.java:117)
at java.util.jar.JarFile.<init>(JarFile.java:55)
<bonniot>and what command do you use?13:10
<arjanb>java -jar -cp nice.jar:nicedoc.jar nice.tools.doc.fun
<bonniot>(yes, the jvm is not very good about such error messages)13:11
don't you use ; on win?
<arjanb>makes no difference
<bonniot>and both jars are in the working dir?13:12
<arjanb>in the same local dir13:13
<bonniot>java -cp /tmp/nicedoc.jar:/usr/share/java/nice.jar nice.tools.doc.fun13:14
works for me
-jar -cp
that's not good
java probably looks for a file called '-cp' ;-/
<arjanb>java -cp nice.jar;nicedoc.jar nice.tools.doc.fun13:16
Exception in thread "main" java.lang.NoSuchMethodError: bossa.modules.Compilation.<init>(ZZZLjava/l
ng/String;Ljava/lang/String;Ljava/lang/String;Ljava/lang/String;ZZLjava/lang/String;Lbossa/modules/
ocator;Lbossa/modules/CompilationListener;Ljava/util/Map;Ljava/util/Map;ZIF)V
at nice.tools.doc.fun.main(~/nicedoc/src/./nice/tools/doc:23)
<bonniot>i think that's using incompatible versions of nicedoc and nice.jar13:17
you should try 0.9.513:18
(is there a new field in Compilation?)
<arjanb>afaik no13:19
with 0.9.5 it runs13:22
Exception in thread "main" java.io.FileNotFoundException: nice\lang\<.html (The filename, directory
name, or volume label syntax is incorrect)
at java.io.FileOutputStream.open(Native Method)
at java.io.FileOutputStream.<init>(FileOutputStream.java:176)
at java.io.FileOutputStream.<init>(FileOutputStream.java:70)
at java.io.FileWriter.<init>(FileWriter.java:46)
at nice.tools.doc.fun.write(~/nicedoc/src/./nice/tools/doc:99)
<CIA-3>03bonniot * 10Nice/ (13 files in 4 dirs): 13:44
Use more typing information from the context when deciding what bytecode types
to use for literal arrays and tuples, especially when they are nested inside
other literal arrays or tuples. This allows to generate more efficient code,
and fixes several runtime errors (fixes #761629 and #892625).
<arjanb>does it pass al tests now?14:06
<bonniot>like yesterday. all execept your last one, which is a different issue as you said14:12
<arjanb>ok14:13
<CIA-3>03bonniot * 10Nice/src/bossa/syntax/InlinedMethod.java: Better print wrapped exceptions.14:14
03bonniot * 10Nice/src/bossa/modules/Package.java: Let the wrapped exception propagate, instead of rewrapping it differently.14:28
* Bluelive joins14:52
<bonniot>hi Bl14:54
<Bluelive>hello15:00
<arjanb>daniel: the development of cecil is still active, at least the research: http://www.cs.washington.edu/research/projects/cecil/www/pubs/vass-thesis.html15:13
<bonniot>saw that, but that thesis had been started a long time ago15:17
what i saw is that the language manual has not been updated since Dev 2002
<arjanb>only a year that's not so strange for an older language15:21
<bonniot>the release is also the same date15:23
just checked the cecil-interest archives, and there is nothing...15:46
so i don't see too much activity...
<CIA-3>03bonniot * 10Nice/Makefile: 15:52
Make the bootstrap work when inlined methods use modified code from the
nice.tools.code package.
<Bluelive>hmz the nosica peeps suggest to drop the try keyword and allow a catch to come after any compound statement15:55
<bonniot>how would that look like?16:04
<Bluelive>{ .. some statements } catch (...) { ... }16:05
altough they immediatly go into the wods by going with something weird
for (...) { ... } else { ... } catch (...) {...}16:06
not a bad idea thoug, losing the try keyword, its basicly a horribly useless thing
in alpha i ditched the required compound statement for similair reasons
<bonniot>well, the weird case shows what try is useful for: show the scope of the catch16:08
so you can have
try foo(); catch() {} ?
<Bluelive>yup16:09
after catch can also be a direct statement16:10
this gives potential for ambiguity like those wiht else when using nested if's with out explicit compounds
but i dont think thats a serious risk/problem16:11
<bonniot>it's the same ambiguity as nested ifs, isn't it?16:24
<Bluelive>yes16:25
maybe something for nice, currently implementing somethig so that you only have to give an initial sourcefile, and the compiler resolves dependancies and issues recompilation of those if their sourcefiles or a dependancie has changed16:55
<bonniot>nice currently does that, only it works with packages instead of source files17:00
i.e. it does not make sense to compile only one source file, you always compile at least one package17:01
why do you say it's for nice, is the implementation generic, or just the idea?
<Bluelive>the idea17:03
alpha doesnt have class files, but im working on something called a bundle
<bonniot>for holding compiled code?
<CIA-3>03bonniot * 10Nice/src/ (3 files in 2 dirs): Minor: more precise typing of the SpecialTypes.array(_) method.17:04
<Bluelive>basicly a fully resolved list of all kinds of things, like all method deescriptions and such, list of dependancies, and maybe in the future a compressed pregenerated c file
currently it recompiles everything, even those that havent changed
<arjanb>i'm trying out haskell and i'm surprised how easy it goes19:17
so Nice is good enough to learn the principles of functional programming19:18
<bonniot>:-)19:51
very nice, but what about line numbers? could it be there for a 0.9.6 tomorrow?19:52
<arjanb>well i'm not sure how to fix that
<bonniot>what's the problem?19:54
<arjanb>catch and print and exit or print and continue or print and rethrow or wrap in other exception19:56
<bonniot>catch,print,exit it dispatch.main seems the simplest19:59
<arjanb>if it's in dispatch.main i have to do bytecode plumbing
<bonniot>the same as fun.main, just at a different location, no?20:00
<arjanb>no in fun.main i can just insert some ast things20:01
<bonniot>i see20:02
i think it's better to do it at the code generation level (gnu.expr) anyway
you can look at the way Try/Catch is compiled, and do the same20:04
* magnus-- joins20:39
<arjanb>hi
<magnus-->hi
my experience with antlr so far: A lot of nice examples, fast to learn. The left factoring is tedious20:41
I think left factoring should be automated20:46
<arjanb>what do you mean with left factoring?20:47
isn't that just a property of all LR based parsers20:58
do you use a default lookahead > 1? antlr is behaving odd with it in some cases21:02
<magnus-->of LL parsers you mean?21:09
<arjanb>right
<magnus-->if I use a >1 lookahead (options{k=3;}) it still gives warnings
because both assignments and variable declarations can start with ID it complains:)21:10
<arjanb>yes but you can use predicates21:11
complete left factoring is very awkward if you want to embed code in the parser21:13
<magnus-->hm yeah, i found something about predicates before i stopped working on it today21:16
I'm still wondering why the parser gen couldn't generate a test automatically to choose the real match, but maybe there are reasons:21:17
.
<arjanb>it all depends on how much tokens the parser looks ahead to before making a choice 21:19
in most parser it's only 1
<magnus-->i see21:34
<arjanb>the main reason to keep it small is that parsing time grows exponentially with the lookahead size21:39
<magnus-->Hmm, even if you build a lookahead tree?21:43
<arjanb>in the worst case yes21:44
<bonniot>typically you set k=1, and you disambiguate with semantic lookahad where needed22:04
<CIA-3>03bonniot * 10swing/project.xml: Added dependency on the maven-nice plugin.23:21

Generated by Sualtam