Re: [gentoo-dev] Subslots progress in main tree

2013-01-24 Thread Ian Stakenvicius
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA256

On 23/01/13 03:42 PM, Tomáš Chvátal wrote:
 Hi guys, do we have some scans that report libraries converted to
 subslots and lists their rdeps checked if they are updated
 accordingly?
 
 It might be pretty usefull to actually see where the deps needed to
 be updated so we can take use of this feature where possible (also
 its a hint for lib maintainers to update their libs and see real
 impact).
 
 Cheers
 
 Tom
 


No, we don't -- and I agree, it would.  I'll look into it.

In general, I would recommend for any library maintainers, to either
file bugs against the rdeps of a package or directly update the rdeps
(with permission, of course) when implementing sub-slots.  Of course,
I believe the rdeps of a library can add slot operators to their
*DEPEND atoms even if the library doesn't have a sub-slot (yet); this
would be useful anyways, as a new sub-slot in the lib will
automatically trigger appropriate rebuilding.  The only issue here is
a maintainer would need to know how sensitive their package is to
changes in the library it deoends on (ie whether it actually needs to
use a := slot operator or not).

-BEGIN PGP SIGNATURE-
Version: GnuPG v2.0.19 (GNU/Linux)

iF4EAREIAAYFAlEBUBMACgkQ2ugaI38ACPDlOgD/THUYZLcO8wT3TRdv/gsEXGsK
aM75M3N/Pu7aFlRd/NIA/05THq9ndnFYO7umS4nYa20E2806B4D8sMEmXoHiUOSc
=35iX
-END PGP SIGNATURE-



Re: [gentoo-dev] Subslots progress in main tree

2013-01-24 Thread Zac Medico
On 01/24/2013 07:15 AM, Ian Stakenvicius wrote:
 In general, I would recommend for any library maintainers, to either
 file bugs against the rdeps of a package or directly update the rdeps
 (with permission, of course) when implementing sub-slots.  Of course,
 I believe the rdeps of a library can add slot operators to their
 *DEPEND atoms even if the library doesn't have a sub-slot (yet); this
 would be useful anyways, as a new sub-slot in the lib will
 automatically trigger appropriate rebuilding.

Yes, that's a good idea. However, it means that the introduction of the
first sub-slot will trigger a rebuild of the dependent package. So, it's
a good idea to introduce the first sub-slot during a version bump where
something has changed that justifies a sub-slot bump.
-- 
Thanks,
Zac



Re: [gentoo-dev] Subslots progress in main tree

2013-01-24 Thread Ian Stakenvicius
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA256

On 24/01/13 11:39 AM, Zac Medico wrote:
 On 01/24/2013 07:15 AM, Ian Stakenvicius wrote:
 In general, I would recommend for any library maintainers, to
 either file bugs against the rdeps of a package or directly
 update the rdeps (with permission, of course) when implementing
 sub-slots.  Of course, I believe the rdeps of a library can add
 slot operators to their *DEPEND atoms even if the library doesn't
 have a sub-slot (yet); this would be useful anyways, as a new
 sub-slot in the lib will automatically trigger appropriate
 rebuilding.
 
 Yes, that's a good idea. However, it means that the introduction of
 the first sub-slot will trigger a rebuild of the dependent package.
 So, it's a good idea to introduce the first sub-slot during a
 version bump where something has changed that justifies a sub-slot
 bump.

That seems to be what dev's are doing anyways, though.
-BEGIN PGP SIGNATURE-
Version: GnuPG v2.0.19 (GNU/Linux)

iF4EAREIAAYFAlEBaK4ACgkQ2ugaI38ACPAAfgD+Ja3RKlfv1uWn7e3/V2rnGNQQ
YY2NHIgQLQxp/Dx66hcA/iPoA7tF6cME8MtqmOumIEVrDzQpE8LEARPmCeQOlqwd
=HzQM
-END PGP SIGNATURE-



Re: [gentoo-dev] Subslots progress in main tree

2013-01-24 Thread Zac Medico
On 01/23/2013 12:42 PM, Tomáš Chvátal wrote:
 It might be pretty usefull to actually see where the deps needed to be
 updated so we can take use of this feature where possible (also its a
 hint for lib maintainers to update their libs and see real impact).

Another useful hint, for those that have FEATURES=preserve-libs enabled,
is the @preserved-rebuild suggestion that emerge shows before it exits
when there are remaining preserved libraries. Any time that you see this
message, it means that there was a missed opportunity for an automatic
rebuild, due to sub-slot and/or slot-operator not being used by the
involved ebuilds.
-- 
Thanks,
Zac



[gentoo-dev] Subslots progress in main tree

2013-01-23 Thread Tomáš Chvátal
Hi guys,
do we have some scans that report libraries converted to subslots and
lists their rdeps checked if they are updated accordingly?

It might be pretty usefull to actually see where the deps needed to be
updated so we can take use of this feature where possible (also its a
hint for lib maintainers to update their libs and see real impact).

Cheers

Tom