Bugzilla Bug 85980

desklets 0.4.6 is out! You may now rename previously ebuild accepted in portage
(see x11-misc/adesklets/adesklets-0.4.5.ebuild) to adesklets-0.4.6.ebuild
without any other change.

It would also be nice if this ebuild has the same masked status as the others -
if gentoo's mainainers decide to keep around adesklets-0.4.4.ebuild or
adesklets-0.4.5.ebuild in portage, they should not have different $KEYWORDS
value than this one... I would rather have everything marked as unstable than
having user not upgrading to the latest revision version. Thanks,

Version 0.4.6 is now in Portage, thank you!

> It would also be nice if this ebuild has the same masked status as the others
> - if gentoo's mainainers decide to keep around adesklets-0.4.4.ebuild or
> adesklets-0.4.5.ebuild in portage, they should not have different $KEYWORDS
> value than this one... I would rather have everything marked as unstable than
> having user not upgrading to the latest revision version.
x86 means stable, ~x86 means testing (same applies to all archs)...
When a new ebuild is added to Portage, it is generally masked for testing until the dev who added it is confident that it will not break anyone's machine. As such, there is no adding a new package version directly as a stable ebuild. After that, when a new version is masked stable, the older version is pruned out from the tree (unless there is some reason to keep it).

So, as you can see, version 0.4.6 is now in Portage (for testing), 0.4.5 just got stable, and 0.4.4 has been removed. Of course, I should only touch keywords I have the hardware for, so there is no way I'm going to mark it stable for amd64 too. It will take some other dev who is interested in the package and has access to an amd64 machine for that.

> As such, there is no adding a new package version directly as a stable ebuild

That's not what I asked you for. Would it be possible to reverse everything to testing? 

KEYWORDS="~x86 ~amd64"

on all ebuilds actually present in portage (adesklets-0.4.5.ebuild and adesklets-0.4.6)?

I use the normal GNU convention for naming packages (NAME-MAJOR.MINOR.REVISION), meaning that there is no new features between 0.4.4, 0.4.5 and 0.4.6, only bug fixes. People should by all means install 0.4.6 instead of 0.4.5 or 0.4.4, as it's only better. I do not mind having it marked as testing, as long as it does cause testing people to miss a bug-fix update.

The day I'll release a 0.5.0, then you kewrods scheme would apply, and I guess that, after a while, it would be normal to revert latest 0.4.x serie as stable while 0.5.0 keep being in test... But I is undubitably your call.

Sorry If I should have filed a new bug; I just didn't think it was appropriate.

>> As such, there is no adding a new package version directly as a stable ebuild
> That's not what I asked you for. Would it be possible to reverse everything to testing?

Uh, sorry I mis-understood you...
Done :-)