[gentoo-pms] The order of functions for upgrade/downgrade

[gentoo-pms] The order of functions for upgrade/downgrade

[Petteri Räty]

[-- Attachment #1: Type: text/plain, Size: 743 bytes --]

http://dev.gentoo.org/~zmedico/portage/doc/ch05.html#package-ebuild-phases

Currently PMS has what <2.1.5 does.

21:05 < Betelgeuse> zmedico: We should put a note to devmanual that if
you rely on order of things you should use at least that EAPI
21:06 < zmedico> Betelgeuse: EAPI 1
21:06 < zmedico> oh, I mean EAPI 2
21:06 < zmedico> <2.1.5 supported EAPI 1

What is the order that other PMs currently use? If the new order is used
by everyone currently maybe we should change the order for EAPIs like 2
and 3. At least should have the current behavior of Portage for EAPI 4
so that if people need it then they should use EAPI 4. I know that
Portage shouldn't have changed it so no reason to bring it up.

Regards,
Petteri


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 900 bytes --]

Re: [gentoo-pms] The order of functions for upgrade/downgrade

[David Leverton]

On Saturday 03 April 2010 22:11:55 Petteri Räty wrote:
> What is the order that other PMs currently use?

Paludis uses the "old" order for EAPIs 0 and 1, and the "new" order for 2 and 
3 (going by the EAPI of the newly-installed package).  I thought the plan was 
for PMS to specify the new order for 2 and later (and in practice, probably 
make it undefined for 0 and 1) so it's probably an oversight that it doesn't.

Re: [gentoo-pms] The order of functions for upgrade/downgrade

[Petteri Räty]

[-- Attachment #1: Type: text/plain, Size: 613 bytes --]

On 04/04/2010 12:19 AM, David Leverton wrote:
> On Saturday 03 April 2010 22:11:55 Petteri Räty wrote:
>> What is the order that other PMs currently use?
> 
> Paludis uses the "old" order for EAPIs 0 and 1, and the "new" order for 2 and 
> 3 (going by the EAPI of the newly-installed package).  I thought the plan was 
> for PMS to specify the new order for 2 and later (and in practice, probably 
> make it undefined for 0 and 1) so it's probably an oversight that it doesn't.
> 

Yeah this seems prudent to me and was actually what we were just talking
about on #gentoo-dev.

Regards,
Petteri


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 900 bytes --]