[Laszlo-dev] Re: Ant 1.6.1

[Mika Göckel]

Actually I didn't bother, because I don't use jikes at all. As the mean 
time in the build process is not spent compiling the java sources (482 
source files in about 30s on my Laptop) but compiling the lfc classes in 
org.openlaszlo.sc.Main... so I don't see, that a faster javac would help 
much.
But anyway, on my 2GHz Pentium-M Laptop, "ant clean build" takes 4 
minutes 18 seconds.... (swf6/7/8).

The ant doc says about jikes, that the system property jikes.class.path 
controlles it (and you need to set the compiler attribute to "jikes"):

|jikes.class.path| 	path 	The specified path is added to the classpath 
if jikes is used as compiler.


(http://ant.apache.org/manual/running.html#sysprops)

My opinion is, that the choice of the compiler should be with the user 
who builds it, as long as we don't have a better reason to stick to a 
specific one.


Scott Evans schrieb:

>Yah, build.compiler sets the compiler to be sued, but as far as I can tell 
>>from reading the ant source, there's no way to tell it where to find the 
>executable short of using that attribute.  And of course the ant docs are 
>a little spotty on this.  Mika, ideas?  How did you deal with the jikes 
>thing in your 1.6.x port?
>
>
>
>gse
>
>
>
>
>On Tue, 22 Nov 2005, Joseph Silverman wrote:
>
>  
>
>>I believe that the build.compiler variable was there in 1.5 but that it
>>wouldn't take a path, or something like that - hence the build.compiler.path
>>variable I added (and code to handle it).
>>
>>On Nov 22, 2005, at 4:06 AM, Oliver Steele wrote:
>>
>>    
>>
>>>Jikes is much faster.  Google says
>>><http://ant.apache.org/manual/CoreTasks/javac.html> that there's a
>>>build.compiler attribute.  Is this page for a different version of ant?
>>>
>>>On Nov 21, 2005, at 11:50 PM, Scott Evans wrote:
>>>
>>>      
>>>
>>>>ah!
>>>>
>>>>But that's one of very few things that's keeping us from using a stock
>>>>ant 1.6.5.  And stock is better, no?
>>>>
>>>>Seems like it'd be better, if we must use jikes, to use the 'executable'
>>>>attribute that was added in 1.6, even if we have to put it on every javac
>>>>command.
>>>>
>>>>
>>>>gse
>>>>
>>>>
>>>>On Mon, 21 Nov 2005, Joseph Silverman wrote:
>>>>
>>>>        
>>>>
>>>>>actually, ant 1.5 in //depot/tools has been modded, by yours truly, to
>>>>>allow a
>>>>>jikes compiler path to be passed into it.  you will have to apply that
>>>>>mod to
>>>>>ant 1.6 to get similar functionality!
>>>>>
>>>>>On Nov 21, 2005, at 6:17 PM, Scott Evans wrote:
>>>>>
>>>>>          
>>>>>
>>>>>>I'm trying to get lps-pave to build with ant 1.6.5 -- I've had to make
>>>>>>very few changes so far but I can't get Jikes to work.  It seems that
>>>>>>1.6.? may determine the jikes executable differently.  Our current
>>>>>>code
>>>>>>does this by setting build.compiler.path -- but I don't see any
>>>>>>reference
>>>>>>to build.compiler.path in the ant source code.
>>>>>>
>>>>>>There's now an "executable" attribute that you can add to all <javac/>
>>>>>>elements, but that seems non-universal and therefore stupid.  It does
>>>>>>work, however.
>>>>>>
>>>>>>Anyone know anything about this?
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>On Sat, 19 Nov 2005, P T Withington wrote:
>>>>>>
>>>>>>            
>>>>>>
>>>>>>>Just getting around to trying this out, because I want to enable
>>>>>>>assertions in
>>>>>>>our .java.
>>>>>>>
>>>>>>>Seems to work fine for me for building.
>>>>>>>
>>>>>>>On 17 Mar 2004, at 14:31, Joseph Silverman wrote:
>>>>>>>
>>>>>>>              
>>>>>>>
>>>>>>>>I installed ant version 1.6.1 in the
>>>>>>>>//depot/tools/apache-ant-1.6.1
>>>>>>>>directory.  I installed, into it, the three .jar files requires to
>>>>>>>>compile
>>>>>>>>LPS.  You can use it by simply setting your ANT_HOME environment
>>>>>>>>to this
>>>>>>>>directory.  You MIGHT need to make one more change, to the
>>>>>>>>CLASSPATH
>>>>>>>>environment variable.  I had to change it to reference
>>>>>>>>$TOMCAT_HOME/lib/server/lib/catalina-ant.jar rather than whatever
>>>>>>>>it was
>>>>>>>>referencing before.
>>>>>>>>
>>>>>>>>Please give this pup a try and get back to me if you have
>>>>>>>>problems.
>>>>>>>>                
>>>>>>>>
>_______________________________________________
>Laszlo-dev mailing list
>Laszlo-dev at openlaszlo.org
>http://www.openlaszlo.org/mailman/listinfo/laszlo-dev
>
>  
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://openlaszlo.org/pipermail/laszlo-dev/attachments/20051122/7e84b1a9/attachment.html