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

Using timezone: Central European Time
<bonniot>good night01:43
* bonniot leaves01:59
* arjanb leaves05:39
* bonniot joins10:06
* arjanb joins10:20
* cm_ joins13:18
hi
<arjanb>hi13:22
indexing like x[a,b] should work now in the cvs version13:23
<CIA-7>03arjanb * 10Nice/src/gnu/ (3 files in 2 dirs): Removed dependency on the gnu.math package.13:43
<cm_>very nice indeed!13:50
Are you planning to release 1.0 soon?15:08
<bonniot>0.9.9 soon15:15
there are a few more features to be implemented before 1.015:16
<cm_>It might be a good idea to review stdlib in order to avoid future combability problems15:19
Is nicedoc ready?15:20
I have some trouble compiling the latest cvs. I am using the latest nice.sf.net/nice.jar15:28
<arjanb>what't the error15:30
?
<cm_>I'm trying make clean. Let see if that works15:31
I get a lot of "Comparing a non-null value with null" though
"make clean" was the problem15:32
<arjanb>that are just warnings
<cm_>Pointless checking?15:33
I suppose15:34
<arjanb>yes though most warnings are caused by imprecise types of java methods used15:35
<bonniot>about nicedoc: it's starting to be usable, but it's still rough at the edges15:43
for stdlib, your comments are welcome ;-)
<cm_>Why is T get(T[],int) defined for primitives? Isn't that a integral part of the language?16:08
arjanb: Do you remember the origin of the problem a had with insert() and StringBuffers?16:10
<bonniot>get for primitives is just an optimization16:18
it can be compiled as a single bytecode
<cm_>So array types are "generic" in the compiler?16:38
Oppesed to Java?16:39
Opposed
bonniot: Are you planning to compile specific instances of generic functions for primitive types for 1.0?16:41
<bonniot>generic in the sense that you can pass an int[] to a function expecting T[], yes16:42
I don't see that kind of specialization as a must for 1.016:43
so it probably won't happen before 1.0, unless somebody else decides to work on it
<cm_>Agree, just for performance. It is a pitty though that you can't create a generic Matrix class due to the performance penalties.16:44
<bonniot>yes16:45
<cm_>Something for post 1.0 maybe
<bonniot>but if it's fully generic you could just store things, not do operations like addition because you don't know the type of the elements16:46
so the part that is reusable is not very big
<cm_>True16:47
The penalty is however apparent for generic functions like map, app etc for arrays. I think a have to stop nagging about this ;-)16:50
The new get functions works great btw
<bonniot>if you mesure that e.g. map has a serious influence on your performance, you could try writing the specialized version by hand17:11
<cm_>Yes, I have written specialized functions.17:12
Got to go. Later
* cm_ leaves
<CIA-7>03arjanb * 10Nice/src/nice/tools/compiler/console/listener.nice: Warning prefix for warnings in the console.18:20
<bonniot>that's a good idea18:24
<CIA-7>03arjanb * 10Nice/src/bossa/ (6 files in 2 dirs): Converted ExpressionStmt to Nice code.20:05
03arjanb * 10Nice/src/gnu/expr/Compilation.java: Silenced superfluous 'cannot convert literal' warnings.20:22

Generated by Sualtam