new module

new module

[Tom Tromey]

I added a new 'builder' module to mauve.
This has the scripts we use on builder.classpath.org to drive the
nightly builds.

builder isn't yet set up to use this code, but hopefully I'll be
doing that a bit later.  First I wanted to clean up a few things in
the scripts.

Tom

Re: new module

[Andrew John Hughes]

[-- Attachment #1: Type: text/plain, Size: 1107 bytes --]

On Mon, 2006-06-19 at 10:00 -0600, Tom Tromey wrote:
> I added a new 'builder' module to mauve.
> This has the scripts we use on builder.classpath.org to drive the
> nightly builds.
> 
> builder isn't yet set up to use this code, but hopefully I'll be
> doing that a bit later.  First I wanted to clean up a few things in
> the scripts.
> 
> Tom

Great news!  Particularly take a look at ecj, which still seems to be
failing, despite me pointing it at our built version in two different
ways... :(
-- 
Andrew :-)

Please avoid sending me Microsoft Office (e.g. Word, PowerPoint)
attachments.
See http://www.fsf.org/philosophy/no-word-attachments.html

If you use Microsoft Office, support movement towards the end of vendor
lock-in:
http://opendocumentfellowship.org/petition/

"Value your freedom, or you will lose it, teaches history.
`Don't bother us with politics' respond those who don't want to learn."
-- Richard Stallman

Escape the Java Trap with GNU Classpath!
http://www.gnu.org/philosophy/java-trap.html
public class gcj extends Freedom implements Java { ... }


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: new module

[Mark Wielaard]

[-- Attachment #1: Type: text/plain, Size: 1039 bytes --]

On Mon, 2006-06-19 at 18:03 +0100, Andrew John Hughes wrote:
> On Mon, 2006-06-19 at 10:00 -0600, Tom Tromey wrote:
> > I added a new 'builder' module to mauve.
> > This has the scripts we use on builder.classpath.org to drive the
> > nightly builds.
> > 
> > builder isn't yet set up to use this code, but hopefully I'll be
> > doing that a bit later.  First I wanted to clean up a few things in
> > the scripts.
> > 
> > Tom
> 
> Great news!  Particularly take a look at ecj, which still seems to be
> failing, despite me pointing it at our built version in two different
> ways... :(

Yes, thanks a lot Tom!
I saw Tom added a patch for real gcjx removal in the builder module. I
have installed that patch now on builder.classpath.org and will monitor
a full build cycle. It seems we were using the gcjx build a lot still
even though I disabled it (in a crappy way, now that I see all the
places Tom needed to patch, sorry about that). If things go right this
should get our ecj-bootstrap back.

Cheers,

Mark

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: new module

[Andrew John Hughes]

On 21 Jun 2006, at 08:52, Mark Wielaard wrote:

> On Mon, 2006-06-19 at 18:03 +0100, Andrew John Hughes wrote:
>> On Mon, 2006-06-19 at 10:00 -0600, Tom Tromey wrote:
>>> I added a new 'builder' module to mauve.
>>> This has the scripts we use on builder.classpath.org to drive the
>>> nightly builds.
>>>
>>> builder isn't yet set up to use this code, but hopefully I'll be
>>> doing that a bit later.  First I wanted to clean up a few things in
>>> the scripts.
>>>
>>> Tom
>>
>> Great news!  Particularly take a look at ecj, which still seems to be
>> failing, despite me pointing it at our built version in two different
>> ways... :(
>
> Yes, thanks a lot Tom!
> I saw Tom added a patch for real gcjx removal in the builder module. I
> have installed that patch now on builder.classpath.org and will  
> monitor
> a full build cycle. It seems we were using the gcjx build a lot still
> even though I disabled it (in a crappy way, now that I see all the
> places Tom needed to patch, sorry about that). If things go right this
> should get our ecj-bootstrap back.
>
> Cheers,
>
> Mark

Hallelujah! We might finally see the result of Tom's concurrency patch.
--
Andrew :)

Postgraduate Student
Verification and Testing Group
Department of Computer Science
University of Sheffield

Re: new module

[Mark Wielaard]

On Wed, 2006-06-21 at 09:45 +0100, Andrew John Hughes wrote:
> Hallelujah! We might finally see the result of Tom's concurrency patch.

:) We might, but apart from one small typo in the Ecj build script,
fixed as follows:

2006-06-21  Mark Wielaard  <mark@klomp.org>

        * Ecj: Fix jikese typo.

there is a compilation error on the generics branch which might or might
not be valid (I don't have the language spec handy):

3. ERROR in ../../classpath/java/util/Collections.java
 (at line 6622)
        (Class<Map.Entry<K,V>>) Map.Entry.class;
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Cannot cast from Class<Map.Entry> to Class<Map.Entry<K,V>>

Cheers,

Mark

--- scripts/Ecj 19 Jun 2006 19:53:33 -0000      1.2
+++ scripts/Ecj 21 Jun 2006 10:48:57 -0000
@@ -88,7 +88,7 @@
 cd org.eclipse.jdt.core
 exec > ../ecj-native-build/Log.Std 2>&1
 set -e
-find compiler batch -name \*.java | CLASSPATH=compiler:batch xargs ../ecj-gcj-build/ecj -1.4 -bootclasspath $TOP/classpath/jikese-install/share/classpath/glibj.zip -classpath compiler:batch -d ../ecj-native-build -nowarn
+find compiler batch -name \*.java | CLASSPATH=compiler:batch xargs ../ecj-gcj-build/ecj -1.4 -bootclasspath $TOP/classpath/jikes-install/share/classpath/glibj.zip -classpath compiler:batch -d ../ecj-native-build -nowarn
 )

 Report "ecj built with native-ecj" $? ecj-native-build/Log.Std

RE: new module

[Jeroen Frijters]

Mark Wielaard wrote:
> On Wed, 2006-06-21 at 09:45 +0100, Andrew John Hughes wrote:
> > Hallelujah! We might finally see the result of Tom's 
> concurrency patch.
> 
> :) We might, but apart from one small typo in the Ecj build script,
> fixed as follows:
> 
> 2006-06-21  Mark Wielaard  <mark@klomp.org>
> 
>         * Ecj: Fix jikese typo.
> 
> there is a compilation error on the generics branch which 
> might or might
> not be valid (I don't have the language spec handy):
> 
> 3. ERROR in ../../classpath/java/util/Collections.java
>  (at line 6622)
>         (Class<Map.Entry<K,V>>) Map.Entry.class;
>         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> Cannot cast from Class<Map.Entry> to Class<Map.Entry<K,V>>

I'm no generics expert, but I've given this a little thought and I think
that ecj is right. I will commit my patch.

Regards,
Jeroen

Re: new module

[Andrew John Hughes]

[-- Attachment #1: Type: text/plain, Size: 2525 bytes --]

On Wed, 2006-06-21 at 12:54 +0200, Mark Wielaard wrote:
> On Wed, 2006-06-21 at 09:45 +0100, Andrew John Hughes wrote:
> > Hallelujah! We might finally see the result of Tom's concurrency patch.
> 
> :) We might, but apart from one small typo in the Ecj build script,
> fixed as follows:
> 
> 2006-06-21  Mark Wielaard  <mark@klomp.org>
> 
>         * Ecj: Fix jikese typo.
> 
> there is a compilation error on the generics branch which might or might
> not be valid (I don't have the language spec handy):
> 
> 3. ERROR in ../../classpath/java/util/Collections.java
>  (at line 6622)
>         (Class<Map.Entry<K,V>>) Map.Entry.class;
>         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> Cannot cast from Class<Map.Entry> to Class<Map.Entry<K,V>>
> 
> Cheers,
> 
> Mark
> 
> --- scripts/Ecj 19 Jun 2006 19:53:33 -0000      1.2
> +++ scripts/Ecj 21 Jun 2006 10:48:57 -0000
> @@ -88,7 +88,7 @@
>  cd org.eclipse.jdt.core
>  exec > ../ecj-native-build/Log.Std 2>&1
>  set -e
> -find compiler batch -name \*.java | CLASSPATH=compiler:batch xargs ../ecj-gcj-build/ecj -1.4 -bootclasspath $TOP/classpath/jikese-install/share/classpath/glibj.zip -classpath compiler:batch -d ../ecj-native-build -nowarn
> +find compiler batch -name \*.java | CLASSPATH=compiler:batch xargs ../ecj-gcj-build/ecj -1.4 -bootclasspath $TOP/classpath/jikes-install/share/classpath/glibj.zip -classpath compiler:batch -d ../ecj-native-build -nowarn
>  )
> 
>  Report "ecj built with native-ecj" $? ecj-native-build/Log.Std
> 
> 

My reading of the language spec. is that ecj is right (and
should have been earlier), but feel free to disagree.

A cast from one parameteric type to another, where both are fully
specified (i.e. no wildcards), is basically an equality test i.e. the
conversion from G(T_1, T_2 ... T_n) to G(S_1, S_2 ... S_n) holds if
S_1 = T_1 etc. where none of T_1..T_n or S_1..S_n are wildcards.

-- 
Andrew :-)

Please avoid sending me Microsoft Office (e.g. Word, PowerPoint)
attachments.
See http://www.fsf.org/philosophy/no-word-attachments.html

If you use Microsoft Office, support movement towards the end of vendor
lock-in:
http://opendocumentfellowship.org/petition/

"Value your freedom, or you will lose it, teaches history.
`Don't bother us with politics' respond those who don't want to learn."
-- Richard Stallman

Escape the Java Trap with GNU Classpath!
http://www.gnu.org/philosophy/java-trap.html
public class gcj extends Freedom implements Java { ... }


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]