Greg Tassone wrote:
> On Sat, 2005-12-24 at 16:34 +0200, Petteri Räty wrote:
> 
>>At the moment we have old versions of at least
>>dev-java/{kaffe,jamvm,sablevm} marked stable. The open source java stack
>>is starting to be usable but these old versions certainly are not drop
>>in replacements for the proprietary ones. This in mind I propose that we
>>move everything to ~arch and re-evaluate them going stable when the time
>>is right. To give everyone time for objections I plan on moving the
>>versions to ~arch in January.
> 
> 
> I think the above statements need some clarification.  Are you saying
> that you want to take the currently-marked-as-stable versions of these
> packages in Portage and change them to ~arch?  If so, that is probably a
> bad idea for several reasons, chief of which is the many
> questions/complaints we will all receive when world updates are trying
> to downgrade packages, or worse, when the new Portage starts complaining
> about a broken state of the world file due to "No packages being
> available for [whatever]".

Yes, you got it right. I want to change KEYWORDS from ~x86 to x86.

> 
> Instead I would suggest leaving the existing flags as-is, and bump revs
> on (new) ebuilds (or newer versions if they exist) and just flag those
> as appropriate.

Well seeing that I haven't gotten anyone agreeing with me, that is what
I should do.

> 
> I think most/all folks using those packages are aware of their limited
> compatibility with the proprietary VM's.  Therefore, the risk of leaving
> the current versions "stable" is probably minimal.

This is probably mostly true, but let's see what eix jamvm says:

betelgeuse@pena /usr/share/doc $ eix jamvm
* dev-java/jamvm
     Available versions:  1.3.0 1.3.1 1.3.3 1.4.1
     Installed:           none
     Homepage:            http://jamvm.sourceforge.net/
     Description:         An extremely small and specification-compliant
virtual machine.

jamvm is of course spefication-compliant so this is true, but
gnu-classpath is far from being compatible with the Sun class library
(1.4).

> 
> As a worst-case, if you're really concerned about users misconstruing
> the supposed "stable" status of these packages, you could always add
> some einfo/ewarn style messages to explain it on those versions.
> 

Here lies the reason of me not liking them being stable. Stable packages
just shouldn't have these einfo or ewarn messages.

Regards,
Petteri