Re: [JPackage-devel] FW: jakarta ant 1.4.1/1.5 ant script

[Nicolas Mailhot <Ni...@laposte.net>]

Le mar 30/04/2002 à 17:42, Stefan Bodewig a écrit :
> On 29 Apr 2002, Nicolas Mailhot <Ni...@laposte.net> wrote:
> > Le lun 29/04/2002 à 12:58, Stefan Bodewig a écrit :
> 
> >> (2) replacing
> >> 
> >> if [...] then
> >>   command
> >> fi
> >> 
> >> with [...] && command
> >> 
> >> While (2) is a question of style (and I prefer ours, YMMV), 
> > 
> > Well, (2) is mostly to shrunk the size of these ifs,
> 
> No strong feelings here, I still prefer indentation to show what's
> going on, that's all.
> 
> >> I'm totally opposed to removing the AIX hack as it would break
> >> backwards compatibilty.  People don't need to set JAVACMD on AIX
> >> right now.
> > 
> > Well, as you've wrote it's a hack and I really don't think we should
> > support all non-standard jvm layouts in the main script (plus it's a
> > hellhole when you have many application scripts to support like we
> > do at jpackege).
> 
> >From a jpackage point of view, you are right.  From an Ant point of
> view, backwards compatibility forces us to keep the hack now and for
> the foreseeable future.
> 
> > so this hack could be shared with tomcat and other java apps.
> 
> fine with me.

Good

> > If core ant jars (i.e the jaxp parser) are loaded in the classpath
> > after the local classpath
> 
> Nit 1: Ant's script puts the jar from ANT_HOME/lib to the front.

Well, I'm afraid the non-jpackage rpm I had on my box didn't. 
 
> > and the local classpath contains saxon jars the style task will bomb
> > because it will detect the Ælfred parser first, and this parser is
> > not supported.
> 
> Nit 2: Ælfred is supported in 1.5.

That's great. However Henri wanted me to test this script with 1.4, and
who can tell if we won't have the same kind of bug later ?

Anyway the new (new) classpath builder is recursive and works with ant
installations in "New ant" and jars like "I love spaces.jar", so please
consider it.

Regards

-- 
Nicolas Mailhot

Re: [JPackage-devel] FW: jakarta ant 1.4.1/1.5 ant script

[Stefan Bodewig <bo...@apache.org>]

On Mon, 6 May 2002, Guillaume Rousse <ro...@ccr.jussieu.fr> wrote:

> However, i 'd prefer to have it called set_cmd or set_command, and
> variable JAVA_CMD or JAVA_COMMAND, for overall coherency.

It's been JAVACMD inside of Ant's script for at least two releases, we
cannot change it (from the external point of view).

Stefan

--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>

Re: [JPackage-devel] FW: jakarta ant 1.4.1/1.5 ant script

[Nicolas Mailhot <Ni...@laposte.net>]

Le lun 06/05/2002 à 14:29, Guillaume Rousse a écrit :
> Ainsi parlait Vendredi 3 Mai 2002 10:41, Stefan Bodewig :
> > On Thu, 2 May 2002, Guillaume Rousse <ro...@ccr.jussieu.fr> wrote:
> > > You have to take cares java environment used matches the java
> > > command (aka you don't want to use ibm sdk 1.3 java with sun sdk 1.4
> > > tools.jar) right ?
> >
> > You can do so right now - but I'm not sure why you'd want to do so 8-)
> > Running Kaffe with Sun's classes?  I really don't know.
> OK, there is a miscomprehension here.
> 
> I was refering to the current behaviour of the set_jvm function to look first 
> a defined JAVA_HOME, then use automatically $JAVA_HOME/bin/java as the 
> command (what i was calling "defining the command from the environment), and 
> fallback to look for java in PATH, then define JAVA_HOME as the grandparent 
> :-) directory of its location (what i was calling "defining the environment 
> from the command"). I just wasn't aware of other jdk layouts, and thought 
> JAVACMD was only a way to define jvm flags.
> 
> So you can forget my previous remark, and a specific function should be added. 
> However, i 'd prefer to have it called set_cmd or set_command, and variable 
> JAVA_CMD or JAVA_COMMAND, for overall coherency.

Well, I did an initial implamentation in jpackage-utils.
Now as long as you can sell your changes to everyone, you're welcome to
build on my work.

Regards,

-- 
Nicolas Mailhot

Re: [JPackage-devel] FW: jakarta ant 1.4.1/1.5 ant script

[Guillaume Rousse <ro...@ccr.jussieu.fr>]

Ainsi parlait Vendredi 3 Mai 2002 10:41, Stefan Bodewig :
> On Thu, 2 May 2002, Guillaume Rousse <ro...@ccr.jussieu.fr> wrote:
> > You have to take cares java environment used matches the java
> > command (aka you don't want to use ibm sdk 1.3 java with sun sdk 1.4
> > tools.jar) right ?
>
> You can do so right now - but I'm not sure why you'd want to do so 8-)
> Running Kaffe with Sun's classes?  I really don't know.
OK, there is a miscomprehension here.

I was refering to the current behaviour of the set_jvm function to look first 
a defined JAVA_HOME, then use automatically $JAVA_HOME/bin/java as the 
command (what i was calling "defining the command from the environment), and 
fallback to look for java in PATH, then define JAVA_HOME as the grandparent 
:-) directory of its location (what i was calling "defining the environment 
from the command"). I just wasn't aware of other jdk layouts, and thought 
JAVACMD was only a way to define jvm flags.

So you can forget my previous remark, and a specific function should be added. 
However, i 'd prefer to have it called set_cmd or set_command, and variable 
JAVA_CMD or JAVA_COMMAND, for overall coherency.
-- 
Guillaume Rousse <ro...@ccr.jussieu.fr>
GPG key http://lis.snv.jussieu.fr/~rousse/gpgkey.html

--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>

Re: [JPackage-devel] FW: jakarta ant 1.4.1/1.5 ant script

[Stefan Bodewig <bo...@apache.org>]

On Thu, 2 May 2002, Guillaume Rousse <ro...@ccr.jussieu.fr> wrote:

> You have to take cares java environment used matches the java
> command (aka you don't want to use ibm sdk 1.3 java with sun sdk 1.4
> tools.jar) right ?

You can do so right now - but I'm not sure why you'd want to do so 8-)
Running Kaffe with Sun's classes?  I really don't know.

Stefan

--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>

Re: [JPackage-devel] FW: jakarta ant 1.4.1/1.5 ant script

[Guillaume Rousse <ro...@ccr.jussieu.fr>]

Ainsi parlait Jeudi 2 Mai 2002 16:22, Stefan Bodewig :
> On 02 May 2002, Nicolas Mailhot <Ni...@laposte.net> wrote:
> > As I wrote in another thread, that's one line in java-functions list
> > of jvms to test if we agree on java-functions use.
>
> Yes, saw it, thanks.
>
> > Could you check the java-function in the jpackage-utils rpm if its
> > suitable for ant commit ?
>
> The only problem I see is set_jvm which would need to be modified to
> deal with JAVACMD as well.  JAVACMD enables Ant users to run a version
> of Java that is different from the one in PATH without setting
> JAVA_HOME at the same time.
You have to take cares java environment used matches the java command (aka you 
don't want to use ibm sdk 1.3 java with sun sdk 1.4 tools.jar) right ?  So, 
what differences makes defining the environment from the command (JAVACMD) , 
or defining the command from environment (JAVA_HOME) ?

> Oh, and JAVA_HOME=`dirname $java`/.. will probably need some quotes
> at least around $java (I'm learning, you see ;-).
Right :-)
-- 
Guillaume Rousse <ro...@ccr.jussieu.fr>
GPG key http://lis.snv.jussieu.fr/~rousse/gpgkey.html

--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>

Re: [JPackage-devel] FW: jakarta ant 1.4.1/1.5 ant script

[Stefan Bodewig <bo...@apache.org>]

On 02 May 2002, Nicolas Mailhot <Ni...@laposte.net> wrote:

> As I wrote in another thread, that's one line in java-functions list
> of jvms to test if we agree on java-functions use.

Yes, saw it, thanks.

> Could you check the java-function in the jpackage-utils rpm if its
> suitable for ant commit ?

The only problem I see is set_jvm which would need to be modified to
deal with JAVACMD as well.  JAVACMD enables Ant users to run a version
of Java that is different from the one in PATH without setting
JAVA_HOME at the same time.

Oh, and JAVA_HOME=`dirname $java`/.. will probably need some quotes
at least around $java (I'm learning, you see ;-).

Stefan

--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>

Re: [JPackage-devel] FW: jakarta ant 1.4.1/1.5 ant script

[Nicolas Mailhot <Ni...@laposte.net>]

Le jeu 02/05/2002 à 11:54, Stefan Bodewig a écrit :
> On 02 May 2002, Nicolas Mailhot <Ni...@laposte.net> wrote:
> > Le jeu 02/05/2002 à 11:33, Stefan Bodewig a écrit :
> >> 
> >> "Ant's" is not and this is for a reason - JAVA_HOME/lib/ext isn't
> >> recursive either.
> > 
> > Ok. If you really think being recursive is bad, that's just a switch
> > in find.
> 
> We should do that.  We also need to check whether it does what we want
> it to do on Cygwin (I cannot test it myself).

Ok, unrecursed classpath builder and jikes osx jikespath

> * You need to merge in Diane's Darwin hack (of course I can do that
> myself, but as you are volunteering 8-)

As I wrote in another thread, that's one line in java-functions list of
jvms to test if we agree on java-functions use.
 
> * The default location for java-functions should be in ANT_HOME/bin,
> not ANT_HOME/lib

Did it

> * we need your version of java-functions to commit it at the same time
> 8-)

Could you check the java-function in the jpackage-utils rpm if its
suitable for ant commit ? (of course the list still misses the OSX jvm
define and the set_javacmd but I can add them quickly if you agree on
the core script)

> * putting tools.jar at the front rather then the end of CLASSPATH at
> least needs some more discussion.  I cannot see any problems with
> this approach, but maybe others do.

I guess if no one manifests himself we'll just have to commit it as is. 

Regards,

-- 
Nicolas Mailhot

Re: [JPackage-devel] FW: jakarta ant 1.4.1/1.5 ant script

[Stefan Bodewig <bo...@apache.org>]

On 02 May 2002, Nicolas Mailhot <Ni...@laposte.net> wrote:
> Le jeu 02/05/2002 à 11:33, Stefan Bodewig a écrit :
>> 
>> "Ant's" is not and this is for a reason - JAVA_HOME/lib/ext isn't
>> recursive either.
> 
> Ok. If you really think being recursive is bad, that's just a switch
> in find.

We should do that.  We also need to check whether it does what we want
it to do on Cygwin (I cannot test it myself).

> However I'm not sure that in ant lib context it wouldn't be a good
> thing.

At least one proposal for Ant's future of task libraries would be
broken if we'd make it recursive.  The idea was to have subdirectories
in ANT_HOME/lib and use different class loaders per subdirectory.  The
last thing you'd want would be that these classes would all end up on
the system classloader as well.

> I will try to solve any sticking points you raise till the whole
> thing ends up in ant cvs:)

Great.

* You need to merge in Diane's Darwin hack (of course I can do that
myself, but as you are volunteering 8-)

* The default location for java-functions should be in ANT_HOME/bin,
not ANT_HOME/lib

* we need your version of java-functions to commit it at the same time
8-)

* putting tools.jar at the front rather then the end of CLASSPATH at
least needs some more discussion.  I cannot see any problems with
this approach, but maybe others do.

Stefan

--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>

Re: [JPackage-devel] FW: jakarta ant 1.4.1/1.5 ant script

[Nicolas Mailhot <Ni...@laposte.net>]

Le jeu 02/05/2002 à 11:33, Stefan Bodewig a écrit :
> On 02 May 2002, Nicolas Mailhot <Ni...@laposte.net> wrote:
> 
> > Anyway the new (new) classpath builder is recursive
> 
> "Ant's" is not and this is for a reason - JAVA_HOME/lib/ext isn't
> recursive either.

Ok. If you really think being recursive is bad, that's just a switch in
find. Just tell me and I'll un-recurse it. However I'm not sure that in
ant lib context it wouldn't be a good thing.

> > and works with ant installations in "New ant" and jars like "I love
> > spaces.jar", so please consider it.
> 
> I will give it a very serious look, thanks for your effort and
> persistence.

I will try to solve any sticking points you raise till the whole thing
ends up in ant cvs:)

Regards, 

-- 
Nicolas Mailhot

Re: [JPackage-devel] FW: jakarta ant 1.4.1/1.5 ant script

[Stefan Bodewig <bo...@apache.org>]

On 02 May 2002, Nicolas Mailhot <Ni...@laposte.net> wrote:

> Anyway the new (new) classpath builder is recursive

"Ant's" is not and this is for a reason - JAVA_HOME/lib/ext isn't
recursive either.

> and works with ant installations in "New ant" and jars like "I love
> spaces.jar", so please consider it.

I will give it a very serious look, thanks for your effort and
persistence.

Stefan

--
To unsubscribe, e-mail:   <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>