| 1 |
-----BEGIN PGP SIGNED MESSAGE----- |
| 2 |
Hash: SHA1 |
| 3 |
|
| 4 |
Petteri Räty wrote: |
| 5 |
> Peter Volkov wrote: |
| 6 |
>> Hi. How do we handle packages that provide client, server, and |
| 7 |
>> possibly extra tools/libraries? Do we split packages like binary |
| 8 |
>> distros do or do we use USE flags? What USE flags? Currently some |
| 9 |
>> packages are split other use client, server or minimal USE |
| 10 |
>> flag(s). |
| 11 |
>> |
| 12 |
>> Back in 2006 similar problem was discussed many times with no |
| 13 |
>> final resolution - it was hard to ban split packages since |
| 14 |
>> portage had no support for USE deps. Also some packages started |
| 15 |
>> to utilize 'minimal' USE flag to force users read USE flag |
| 16 |
>> description and thus reduce its usage and lower number of bugs |
| 17 |
>> due to not-installed parts of package. |
| 18 |
>> |
| 19 |
>> With EAPI=2 both use deps and USE defaults (if necessary) are |
| 20 |
>> here so it's possible to introduce some guidelines: |
| 21 |
>> |
| 22 |
>> 1. do not split packages; use USE flags and USE deps. 2. stop |
| 23 |
>> using minimal USE flag to build client or sever only. |
| 24 |
>> |
| 25 |
>> |
| 26 |
>> So are there any good reasons to split packages? |
| 27 |
>> |
| 28 |
>> |
| 29 |
>> https://bugs.gentoo.org/12499 but many similar disscussions were |
| 30 |
>> on this list... |
| 31 |
>> |
| 32 |
> |
| 33 |
> I think a good guideline is: 1. Use a single pkg when upstream |
| 34 |
> releases server and client in one bundle |
| 35 |
|
| 36 |
As always, there are exceptions. I think most users agree that the |
| 37 |
decision of the KDE team to move from the 13 tarballs upstream |
| 38 |
releases to the ~300 ebuilds we have in Gentoo was a good move. From |
| 39 |
the maintainers POV there can be no doubt about it. |
| 40 |
I know KDE doesn't provide servers + clients, but I think it's |
| 41 |
Gentoo's extreme case of split ebuilds. |
| 42 |
|
| 43 |
> 2. Use separate packages when upstream releases client and server |
| 44 |
> separately |
| 45 |
> |
| 46 |
> I think the minimal use flag should not be used for this purpose |
| 47 |
> any more. |
| 48 |
> |
| 49 |
> Regards, Petteri |
| 50 |
> |
| 51 |
I also would like to recall the old discussion about the "usefulness" |
| 52 |
or "correctness" of the "client" and "server" use flags. IIRC, the |
| 53 |
handbook should still list those as not appropriate for use flags, |
| 54 |
given the myriad of different meanings they can carry - even though |
| 55 |
use.local.desc can help here. |
| 56 |
I wonder how many users would like to see mysql move to a split model |
| 57 |
(no criticism intended to anyone that has worked on it over the years) |
| 58 |
or what users feel about the split done on postgresql. Looking at the |
| 59 |
ebuilds (I've been using mysql for years and never used postgresql), |
| 60 |
this is one case where upstream releases a single tarball and one team |
| 61 |
moved to split ebuilds and the other keeps using monos. It would be |
| 62 |
interesting to hear the maintainers' opinion. |
| 63 |
|
| 64 |
- -- |
| 65 |
Regards, |
| 66 |
|
| 67 |
Jorge Vicetto (jmbsvicetto) - jmbsvicetto at gentoo dot org |
| 68 |
Gentoo- forums / Userrel / Devrel / SPARC / KDE |
| 69 |
-----BEGIN PGP SIGNATURE----- |
| 70 |
Version: GnuPG v2.0.13 (GNU/Linux) |
| 71 |
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ |
| 72 |
|
| 73 |
iEYEARECAAYFAksADmsACgkQcAWygvVEyAK5CwCdG5GCExo2Pt/rTqwTQhXzCmJ6 |
| 74 |
M9wAoJwGh6UPr0J3hnCppj/bBaP1Tlix |
| 75 |
=eoNk |
| 76 |
-----END PGP SIGNATURE----- |
Archives