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

Using timezone: Central European Time
* CIA-3 leaves05:08
* CIA-3 joins
* CIA-3 leaves06:33
* CIA-3 joins
* CIA-3 leaves06:53
* CIA-3 joins
* bonniot joins11:06
* CIA-3 leaves12:13
* CIA-3 joins
* CIA-3 leaves12:22
* CIA-7 joins12:23
* arjanb joins12:30
hi
<bonniot>hi12:33
does the search box on the main page look fine in IE?12:36
<arjanb>yes12:37
<bonniot>cool12:38
<arjanb>the only difference between IE and firefox i see is that IE makes the menu a little bigger12:40
<bonniot>ok
<CIA-7>03bonniot * 10Nice/web/new.xsl: Added site search box.12:47
<arjanb>what do you think of introducing a sealed modifier as in dylan?12:49
<bonniot>for methods or classes?12:50
<arjanb>both
for classes it makes possible what Kannan described on the wiki12:52
for methods it could solve http://sourceforge.net/tracker/index.php?func=detail&aid=791630&group_id=12788&atid=36278812:53
<bonniot>cannot enums have fields?12:55
<arjanb>enums can have fields but all elements have the same fields then12:57
and the fields are final
<bonniot>ok, so that could be extended12:58
what do you mean by sealed class?12:59
<arjanb>a sealed class cannot be extended outside its package
<bonniot>then it's only a syntactical difference with enums13:00
<arjanb>no enums are group of instances13:01
<bonniot>not if they can have distinct fields13:02
<arjanb>enums elements are singleton so for enum Color there is only 1 red13:04
<bonniot>atm yes. but not if they can have distinct fields
<arjanb>then it isn't an enum anymore13:06
<bonniot>that's only a question of name13:07
<arjanb>if enums element get fields and aren't singleton anymore i'd rather trow them out13:14
<bonniot>why?13:15
i answered to kannan13:19
<arjanb>because you have 2 styles of writing such things without it makes a real difference
<bonniot>classes and enums?13:20
<arjanb>yes
<bonniot>that's exactly the same as classes and sealed classes
<arjanb>i prefer sealed over enums for such things13:23
<bonniot>i don't see the difference, appart from the name
<arjanb>sealed is more general13:24
<bonniot>oh, and the syntax. enums are listed together, while sealed classes would all be at toplevel I guess
<arjanb>yes
<bonniot>how is it more general?
<arjanb>i meant more general in placing them13:26
<bonniot>ok
<arjanb>further the concept of sealed is usefull for methods too
<bonniot>how would a sealed method work when a new subclass is defined in another package?13:27
<arjanb>the coverage test for sealed methods is different so that no subclassing in another package can break the completeness13:29
<bonniot>ok. and why are sealed methods useful?13:30
<arjanb>see that RFE and it could seperate the extendability and visibility of methods and it allows certain optimizations13:33
<bonniot>but that wouldn't solve the RFE unless all methods were sealed13:40
see my answer on the wiki for a better solution
<arjanb>*away for half an hour13:57
* bonniot leaves13:59
* bonniot joins16:20
* arjanb leaves
* arjanb joins18:11
* CIA-7 leaves19:24
* CIA-7 joins
<arjanb>another difference between enums and sealed is that an enum has a flat structure while sealed allows subclasses21:20
the main reasons for adding enums are java competability and the amount of boilplate code you need to ensure singletonness of the elements, both of them don't apply to enum elements with fields21:22
maybe it's better to discuss this later and/or with others21:23
<bonniot>right, this is not urgent21:24
i thought about a simple way to implement constructors with field assignments21:26
<arjanb>how?21:29
<bonniot>look at all the fields declared in the class. add corresponding symbols in the scope, initially without a value, so the current system will check that they are not used before being set21:30
<arjanb>i see21:34
and for fields with a default value?21:35
<bonniot>those can have their default value21:38
<arjanb>except for default values depending on other fields21:40
the easy solution is to require that values are provided for them inside the constructor21:42
<bonniot>yes. atm those are not allowed anyway, right? so we can see about that in a separate stage21:53
<arjanb>indeed22:06

Generated by Sualtam