[aur-general] Java Dependency and Cross-Compilation Questions

[Aaron Schaefer]

So, my new machine is up and running (and I figured out my previous
packaging issues!)...so I'm updating my jGnash package
(http://www.archlinux.org/packages/community/i686/jgnash/) to the
latest release and there is also currently a bug report on the package
(http://bugs.archlinux.org/task/16665). The bug report correctly
states that jGnash will not run with openjdk6 (jre works just fine),
so what is the current policy for handling that fact?

I know that no other packages depend on jre directly, and the prefered
method is now java-runtime, but doesn't that mean that openjdk6 users
will just have this software silently fail? Also, if you're building
an i386 package on an x86_64 machine, is there an easy way to test the
software to make sure that it's actually working on i386? Thanks in
advance...

--
Aaron "ElasticDog" Schaefer

[aur-general] Orphan Packages

[William Díaz]

I changed of distribution, please adopt the packages.

amsn-extras-svn 
archdark-gdm-theme 
broadcom-wl 
cnijfilter-common-old 
colorschemes-kde4 
cpyrit-cuda 
cwrapper 
firefox-nightly 
go-openoffice-dark-gtk-fix 
gssdp 
gstreamer0.10-farsight 
gtk2-theme-mira 
gupnp 
gupnp-igd 
hydroxygen-iconset 
kde4icons-crystaldiamond 
lib32-libsasl 
openbox-xfzen-themes 
oxygenrefit2-icon-theme 
plasma-themes 
pyrit 
scangearmp-mp140 
slimrat-svn 
snack 
vicious-git 
vim-colorschemes 

Re: [aur-general] Java Dependency and Cross-Compilation Questions

[Smartboy]

On Thu, Nov 5, 2009 at 4:24 PM, Aaron Schaefer  wrote:

> So, my new machine is up and running (and I figured out my previous
> packaging issues!)...so I'm updating my jGnash package
> (http://www.archlinux.org/packages/community/i686/jgnash/) to the
> latest release and there is also currently a bug report on the package
> (http://bugs.archlinux.org/task/16665). The bug report correctly
> states that jGnash will not run with openjdk6 (jre works just fine),
> so what is the current policy for handling that fact?
>
> I know that no other packages depend on jre directly, and the prefered
> method is now java-runtime, but doesn't that mean that openjdk6 users
> will just have this software silently fail? Also, if you're building
> an i386 package on an x86_64 machine, is there an easy way to test the
> software to make sure that it's actually working on i386? Thanks in
> advance...
>
> --
> Aaron "ElasticDog" Schaefer
>

I don't know if it is still true (I haven't tested with the latest release,
but I'm pretty sure it is still a problem), but Aptana also has problems
with OpenJDK6. It installs just fine with it, but running it is a whole
different story, and bugs pop out everywhere when using OpenJDK6. Would it
be appropriate to make it depend on JRE instead?

Smartboy

Re: [aur-general] Java Dependency and Cross-Compilation Questions

[Eric Bélanger]

On Thu, Nov 5, 2009 at 7:24 PM, Aaron Schaefer  wrote:
> So, my new machine is up and running (and I figured out my previous
> packaging issues!)...so I'm updating my jGnash package
> (http://www.archlinux.org/packages/community/i686/jgnash/) to the
> latest release and there is also currently a bug report on the package
> (http://bugs.archlinux.org/task/16665). The bug report correctly
> states that jGnash will not run with openjdk6 (jre works just fine),
> so what is the current policy for handling that fact?
>
> I know that no other packages depend on jre directly, and the prefered
> method is now java-runtime, but doesn't that mean that openjdk6 users
> will just have this software silently fail?

In this case, make it depends on jre.  You could put a note in the
PKGBUILD to explain this dependency.  And, when either openjdk or
jgnash  release new versions, you could test to see if they work fine
together so you could switch back the depends to java-runtime.


Also, if you're building
> an i386 package on an x86_64 machine, is there an easy way to test the
> software to make sure that it's actually working on i386? Thanks in
> advance...
>

you could setup a i686 chroot on your x86_64 system.  I believe
there's info in the wiki.

Eric

> --
> Aaron "ElasticDog" Schaefer
>

Re: [aur-general] Java Dependency and Cross-Compilation Questions

[doorknob60]

Here's how I installed my chroot, it works very well (I don't even use any
lib32 or bin32 packages anymore, just this):
http://wiki.archlinux.org/index.php/Arch64_Install_bundled_32bit_system

On Thu, Nov 5, 2009 at 9:29 PM, Eric Bélanger wrote:

> On Thu, Nov 5, 2009 at 7:24 PM, Aaron Schaefer 
> wrote:
> > So, my new machine is up and running (and I figured out my previous
> > packaging issues!)...so I'm updating my jGnash package
> > (http://www.archlinux.org/packages/community/i686/jgnash/) to the
> > latest release and there is also currently a bug report on the package
> > (http://bugs.archlinux.org/task/16665). The bug report correctly
> > states that jGnash will not run with openjdk6 (jre works just fine),
> > so what is the current policy for handling that fact?
> >
> > I know that no other packages depend on jre directly, and the prefered
> > method is now java-runtime, but doesn't that mean that openjdk6 users
> > will just have this software silently fail?
>
> In this case, make it depends on jre.  You could put a note in the
> PKGBUILD to explain this dependency.  And, when either openjdk or
> jgnash  release new versions, you could test to see if they work fine
> together so you could switch back the depends to java-runtime.
>
>
> Also, if you're building
> > an i386 package on an x86_64 machine, is there an easy way to test the
> > software to make sure that it's actually working on i386? Thanks in
> > advance...
> >
>
> you could setup a i686 chroot on your x86_64 system.  I believe
> there's info in the wiki.
>
> Eric
>
> > --
> > Aaron "ElasticDog" Schaefer
> >
>

Re: [aur-general] Java Dependency and Cross-Compilation Questions

[Eric Bélanger]

On Fri, Nov 6, 2009 at 12:35 AM, doorknob60  wrote:
> Here's how I installed my chroot, it works very well (I don't even use any
> lib32 or bin32 packages anymore, just this):
> http://wiki.archlinux.org/index.php/Arch64_Install_bundled_32bit_system
>

Or you can use the official mkarchroot tool:
http://wiki.archlinux.org/index.php/DeveloperWiki:Building_in_a_Clean_Chroot


> On Thu, Nov 5, 2009 at 9:29 PM, Eric Bélanger wrote:
>
>> On Thu, Nov 5, 2009 at 7:24 PM, Aaron Schaefer 
>> wrote:
>> > So, my new machine is up and running (and I figured out my previous
>> > packaging issues!)...so I'm updating my jGnash package
>> > (http://www.archlinux.org/packages/community/i686/jgnash/) to the
>> > latest release and there is also currently a bug report on the package
>> > (http://bugs.archlinux.org/task/16665). The bug report correctly
>> > states that jGnash will not run with openjdk6 (jre works just fine),
>> > so what is the current policy for handling that fact?
>> >
>> > I know that no other packages depend on jre directly, and the prefered
>> > method is now java-runtime, but doesn't that mean that openjdk6 users
>> > will just have this software silently fail?
>>
>> In this case, make it depends on jre.  You could put a note in the
>> PKGBUILD to explain this dependency.  And, when either openjdk or
>> jgnash  release new versions, you could test to see if they work fine
>> together so you could switch back the depends to java-runtime.
>>
>>
>> Also, if you're building
>> > an i386 package on an x86_64 machine, is there an easy way to test the
>> > software to make sure that it's actually working on i386? Thanks in
>> > advance...
>> >
>>
>> you could setup a i686 chroot on your x86_64 system.  I believe
>> there's info in the wiki.
>>
>> Eric
>>
>> > --
>> > Aaron "ElasticDog" Schaefer
>> >
>>
>

Re: [aur-general] Java Dependency and Cross-Compilation Questions

[Heiko Baums]

Am Fri, 6 Nov 2009 00:29:49 -0500
schrieb Eric Bélanger :

> > I know that no other packages depend on jre directly, and the
> > prefered method is now java-runtime, but doesn't that mean that
> > openjdk6 users will just have this software silently fail?
> 
> In this case, make it depends on jre.  You could put a note in the
> PKGBUILD to explain this dependency.  And, when either openjdk or
> jgnash  release new versions, you could test to see if they work fine
> together so you could switch back the depends to java-runtime.

It would be better to set the dependency to java-runtime instead of
jre so that openjdk6 users can install and test jgnash. If jgnash
really fails to run with openjdk6 - I hadn't had any problems with
openjdk6, yet - then a bug report should be filed to upstream of
openjdk6 and/or jgnash.

I don't think it's the matter of downstream to restrict the dependency
and to force the user using jre instead of openjdk6.

So, please, change the dependency to java-runtime.

Heiko

Re: [aur-general] Java Dependency and Cross-Compilation Questions

[Heiko Baums]

Am Fri, 6 Nov 2009 00:45:06 -0500
schrieb Eric Bélanger :

> >> In this case, make it depends on jre.  You could put a note in the
> >> PKGBUILD to explain this dependency.

Btw., putting a note about a restricted dependency in the PKGBUILD is
not the recommended way because the average user doesn't read the
PKGBUILD if a binary package can't be installed due to a wrong
dependency.

Heiko

Re: [aur-general] Java Dependency and Cross-Compilation Questions

[Eric Bélanger]

On Fri, Nov 6, 2009 at 1:21 AM, Heiko Baums  wrote:
> Am Fri, 6 Nov 2009 00:29:49 -0500
> schrieb Eric Bélanger :
>
>> > I know that no other packages depend on jre directly, and the
>> > prefered method is now java-runtime, but doesn't that mean that
>> > openjdk6 users will just have this software silently fail?
>>
>> In this case, make it depends on jre.  You could put a note in the
>> PKGBUILD to explain this dependency.  And, when either openjdk or
>> jgnash  release new versions, you could test to see if they work fine
>> together so you could switch back the depends to java-runtime.
>
> It would be better to set the dependency to java-runtime instead of
> jre so that openjdk6 users can install and test jgnash. If jgnash
> really fails to run with openjdk6 - I hadn't had any problems with
> openjdk6, yet - then a bug report should be filed to upstream of
> openjdk6 and/or jgnash.
>
> I don't think it's the matter of downstream to restrict the dependency
> and to force the user using jre instead of openjdk6.
>
> So, please, change the dependency to java-runtime.
>
> Heiko
>

Yes, these problems should be reported upstream to openjdk6 and/or jgnash.

If you change the depends to java-runtime, it might be a good idea to
display a warning via a post-upgrade message that there might be
problems with openjdk6.  So that users won't open bug reports about
it.

Re: [aur-general] Java Dependency and Cross-Compilation Questions

[Eric Bélanger]

On Fri, Nov 6, 2009 at 1:34 AM, Heiko Baums  wrote:
> Am Fri, 6 Nov 2009 00:45:06 -0500
> schrieb Eric Bélanger :
>
>> >> In this case, make it depends on jre.  You could put a note in the
>> >> PKGBUILD to explain this dependency.
>
> Btw., putting a note about a restricted dependency in the PKGBUILD is
> not the recommended way because the average user doesn't read the
> PKGBUILD if a binary package can't be installed due to a wrong
> dependency.
>
> Heiko
>

That proposed note was not intended for users but for other TU or devs
who might be tempted to "fix" the dependency by changing it to
java-runtime.

Re: [aur-general] Java Dependency and Cross-Compilation Questions

[Heiko Baums]

Am Fri, 6 Nov 2009 02:01:36 -0500
schrieb Eric Bélanger :

> Yes, these problems should be reported upstream to openjdk6 and/or
> jgnash.
> 
> If you change the depends to java-runtime, it might be a good idea to
> display a warning via a post-upgrade message that there might be
> problems with openjdk6.  So that users won't open bug reports about
> it.

This is the better way of handling this.

If you know of a already filed bug report then it would probably be a
good idea to add the URL to this bug or the bug tracker and the bug id
to this post-upgrade message, too.

Heiko