All of lore.kernel.org
 help / color / mirror / Atom feed
* RFC: merging recipes from Jalimo
@ 2007-12-11 18:42 Robert Schuster
  2007-12-12 13:29 ` Leon Woestenberg
  2007-12-19 13:54 ` Robert Schuster
  0 siblings, 2 replies; 6+ messages in thread
From: Robert Schuster @ 2007-12-11 18:42 UTC (permalink / raw)
  To: openembedded-devel

[-- Attachment #1: Type: text/plain, Size: 388 bytes --]

Hi,
since I've gotten r/w access to the repo (thanks again!) I have begun to
slowly merge the Jalimo recipes. It all starts with this one:

http://bugs.openembedded.org/show_bug.cgi?id=3457

The recipe that builts the initial Java5-capable Java compiler.

Please comment on the bug and tell me whether it should go in or what
you would like to see changed.

Regards
Robert


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 260 bytes --]

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: RFC: merging recipes from Jalimo
  2007-12-11 18:42 RFC: merging recipes from Jalimo Robert Schuster
@ 2007-12-12 13:29 ` Leon Woestenberg
  2007-12-12 16:01   ` Robert Schuster
  2007-12-19 13:54 ` Robert Schuster
  1 sibling, 1 reply; 6+ messages in thread
From: Leon Woestenberg @ 2007-12-12 13:29 UTC (permalink / raw)
  To: openembedded-devel

Hello Robert,

welcome aboard. A remark from my side, although I do not use java much:


I see the compile task assumes the availability of "javac".

javac -d build -source 1.4 -sourcepath source `cat $list`;

How do you expect this to be available on the host? Will any javac
that's out there be good enough to use?

Currently, OpenEmbedded tries to rely on a defined set of host tools
to be installed, and my concern is that we now must add one (out of a
set of acceptable) java compiler to that external dependency list. If
so, we must do so carefully.

Regards,

Leon Woestenberg.



^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: RFC: merging recipes from Jalimo
  2007-12-12 13:29 ` Leon Woestenberg
@ 2007-12-12 16:01   ` Robert Schuster
  2007-12-12 21:13     ` Paul Sokolovsky
  0 siblings, 1 reply; 6+ messages in thread
From: Robert Schuster @ 2007-12-12 16:01 UTC (permalink / raw)
  To: openembedded-devel

[-- Attachment #1: Type: text/plain, Size: 1852 bytes --]

Hi Leon,

Leon Woestenberg schrieb:
> Hello Robert,
> I see the compile task assumes the availability of "javac".
> 
> javac -d build -source 1.4 -sourcepath source `cat $list`;
> 
> How do you expect this to be available on the host? Will any javac
> that's out there be good enough to use?
A Java 1.4 compatible compiler is needed. That can be the proprietary
javac, gcj -C, jikes and ecj. Most distribution have some way to let the
latter three be available as 'javac'.

You may say that we could use jikes-native from OE then but that
compiler needs (as all Java compilers) a class library. Jikes is not
Java5-capable but GNU Classpath expects one since version 0.95. One
could build an older GNU Classpath release just for this purpose but
then there would still be no runtime to use ecj-bootstrap (which is a
Java app itself). Additionally I wanted to stick with latest and
greatest GNU Classpath (especially everything after the Java5 switch)
releases since the older ones contain surprising bugs and miss important
tools (gjar, gjavah etc).

It is really a messy bootstrapping situation, I know. Since OE expects
GCC being installed I think that gij and gcj can also be available on
those systems.

> Currently, OpenEmbedded tries to rely on a defined set of host tools
> to be installed, and my concern is that we now must add one (out of a
> set of acceptable) java compiler to that external dependency list. If
> so, we must do so carefully.
Yes, thats true. A java compiler and a runtime (both from the fairly old
1.4 times) are needed in the beginning. After ecj-bootstrap is built its
only purpose is to built OE's own java runtime and compiler.

If independence from those two pieces of software is important I could
look into this. However I expect this to become complicated.

Regards
Robert


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 260 bytes --]

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: RFC: merging recipes from Jalimo
  2007-12-12 16:01   ` Robert Schuster
@ 2007-12-12 21:13     ` Paul Sokolovsky
  2007-12-15 13:30       ` Leon Woestenberg
  0 siblings, 1 reply; 6+ messages in thread
From: Paul Sokolovsky @ 2007-12-12 21:13 UTC (permalink / raw)
  To: Robert Schuster; +Cc: openembedded-devel

Hello Robert,

Wednesday, December 12, 2007, 6:01:41 PM, you wrote:

> Hi Leon,

> Leon Woestenberg schrieb:
>> Hello Robert,
>> I see the compile task assumes the availability of "javac".
>> 
>> javac -d build -source 1.4 -sourcepath source `cat $list`;
>> 
>> How do you expect this to be available on the host? Will any javac
>> that's out there be good enough to use?
> A Java 1.4 compatible compiler is needed. That can be the proprietary
> javac, gcj -C, jikes and ecj. Most distribution have some way to let the
> latter three be available as 'javac'.

> You may say that we could use jikes-native from OE then but that
> compiler needs (as all Java compilers) a class library. Jikes is not
> Java5-capable but GNU Classpath expects one since version 0.95. One
> could build an older GNU Classpath release just for this purpose but
> then there would still be no runtime to use ecj-bootstrap (which is a
> Java app itself). Additionally I wanted to stick with latest and
> greatest GNU Classpath (especially everything after the Java5 switch)
> releases since the older ones contain surprising bugs and miss important
> tools (gjar, gjavah etc).

> It is really a messy bootstrapping situation, I know. Since OE expects
> GCC being installed I think that gij and gcj can also be available on
> those systems.

>> Currently, OpenEmbedded tries to rely on a defined set of host tools
>> to be installed, and my concern is that we now must add one (out of a
>> set of acceptable) java compiler to that external dependency list. If
>> so, we must do so carefully.
> Yes, thats true. A java compiler and a runtime (both from the fairly old
> 1.4 times) are needed in the beginning. After ecj-bootstrap is built its
> only purpose is to built OE's own java runtime and compiler.

> If independence from those two pieces of software is important I could
> look into this. However I expect this to become complicated.

  As was discussed lately on this list, we should track the expected
host dependencies. This is done using ASSUME_PROVIDED in
conf/bitbake.conf. So, based on the above, I think you want to add
javac-native and java-runtime-native there. Then, your recipe should
DEPENDS on them, and in comment, please include the relevant
additional info above. This should be good compromise, as indeed,
complete self-bootstrapping would be a task in itself, sidetracking you
from the current work.


> Regards
> Robert




-- 
Best regards,
 Paul                            mailto:pmiscml@gmail.com




^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: RFC: merging recipes from Jalimo
  2007-12-12 21:13     ` Paul Sokolovsky
@ 2007-12-15 13:30       ` Leon Woestenberg
  0 siblings, 0 replies; 6+ messages in thread
From: Leon Woestenberg @ 2007-12-15 13:30 UTC (permalink / raw)
  To: openembedded-devel

Hello,

On Dec 12, 2007 10:13 PM, Paul Sokolovsky <pmiscml@gmail.com> wrote:
> Wednesday, December 12, 2007, 6:01:41 PM, you wrote:
> > Hi Leon,
> > Leon Woestenberg schrieb:
> >> Hello Robert,
> >> I see the compile task assumes the availability of "javac".
> >>
> >> javac -d build -source 1.4 -sourcepath source `cat $list`;
> >>
> >> How do you expect this to be available on the host? Will any javac
> >> that's out there be good enough to use?
>
>   As was discussed lately on this list, we should track the expected
> host dependencies. This is done using ASSUME_PROVIDED in
> conf/bitbake.conf. So, based on the above, I think you want to add
> javac-native and java-runtime-native there. Then, your recipe should
> DEPENDS on them, and in comment, please include the relevant
> additional info above. This should be good compromise, as indeed,
> complete self-bootstrapping would be a task in itself, sidetracking you
> from the current work.
>

Yes, fully agreeing with this approach and reasoning.

BTW, I didn't want to sidetrack you, on the contrary, just wanted to
know what the implications where and how we should neatly solve them.

Regards,
-- 
Leon



^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: RFC: merging recipes from Jalimo
  2007-12-11 18:42 RFC: merging recipes from Jalimo Robert Schuster
  2007-12-12 13:29 ` Leon Woestenberg
@ 2007-12-19 13:54 ` Robert Schuster
  1 sibling, 0 replies; 6+ messages in thread
From: Robert Schuster @ 2007-12-19 13:54 UTC (permalink / raw)
  To: openembedded-devel

[-- Attachment #1: Type: text/plain, Size: 1236 bytes --]

Hi,
thanks for all the information. Contrary to what I said in other posts I
will look in providing a completely self-hosting java toolchain. This
will include building one additional instance of classpath and cacao
only (this does not takes as long as the gcc+glibc combo :) ).

The reason for not going the host-provided java route is that I find it
troublesome to add logic to the recipe to find the java's class library
in case it is host provided or self-built.

Regards
Robert

Robert Schuster schrieb:
> Hi,
> since I've gotten r/w access to the repo (thanks again!) I have begun to
> slowly merge the Jalimo recipes. It all starts with this one:
> 
> http://bugs.openembedded.org/show_bug.cgi?id=3457
> 
> The recipe that builts the initial Java5-capable Java compiler.
> 
> Please comment on the bug and tell me whether it should go in or what
> you would like to see changed.
> 
> Regards
> Robert
> 
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 260 bytes --]

^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2007-12-19 14:00 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2007-12-11 18:42 RFC: merging recipes from Jalimo Robert Schuster
2007-12-12 13:29 ` Leon Woestenberg
2007-12-12 16:01   ` Robert Schuster
2007-12-12 21:13     ` Paul Sokolovsky
2007-12-15 13:30       ` Leon Woestenberg
2007-12-19 13:54 ` Robert Schuster

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.