1 |
On Thu, 10 Mar 2022 12:07:40 -0600 |
2 |
William Hubbs <williamh@g.o> wrote: |
3 |
|
4 |
> On Thu, Mar 10, 2022 at 09:29:59AM -0800, Matt Turner wrote: |
5 |
> > On Wed, Mar 9, 2022 at 11:09 PM Joonas Niilola <juippis@g.o> |
6 |
> > wrote: |
7 |
> > > |
8 |
> > > On 9.3.2022 23.00, Matt Turner wrote: |
9 |
> > > > I'd like to deprecate and ultimately remove repoman. I believe |
10 |
> > > > that dev-util/pkgcheck and pkgcommit (from |
11 |
> > > > app-portage/mgorny-dev-scripts) are far superior replacements, |
12 |
> > > > and it makes sense to have people using the same tool and |
13 |
> > > > seeing the same warnings as in the CI. |
14 |
> > > > |
15 |
> > > > Are there any useful checks or behaviors of repoman that are |
16 |
> > > > missing from pkgcheck and pkgcommit? |
17 |
> > > > |
18 |
> > > > Thanks, |
19 |
> > > > Matt |
20 |
> > > > |
21 |
> > > |
22 |
> > > I still fail to see the "why" in here. Repoman is better than |
23 |
> > > pure 'git commit' that I know some people still like to use, and |
24 |
> > > as long as it's kept maintained. |
25 |
> > |
26 |
> > repoman is inferior to other tooling mentioned. The other tooling is |
27 |
> > actually run in CI. Developers should get the same warnings locally |
28 |
> > as in CI. Restated another way: I'm tired of telling people to stop |
29 |
> > using repoman or "pkgcheck would have caught that". |
30 |
> |
31 |
> I am going to nit-pick here, but pkgcheck pulls in pkgcore still. As |
32 |
> far asI know, pkgcore was meant to be a portage-like package manager, |
33 |
> but it isn't maintained. So, can we break that dependency before we |
34 |
> make pkgcheck the official tool for qa checks? I would rather not have |
35 |
> pkgcore landing on everyone's systems unless it is usable. If I am |
36 |
> wrong about pkgcore, please correct me and I'll be quiet, but if not |
37 |
> let's make pkgcheck independent from it before we deprecate repoman. |
38 |
> |
39 |
> Thanks, |
40 |
> |
41 |
> William |
42 |
|
43 |
William. pkgcheck uses pkgcore base code as it's library for accessing |
44 |
and evaluating repositories and ebuilds just like repoman uses portage |
45 |
api's to do the same. This is why pkgcheck is primarily faster than |
46 |
repoman. pkgcore was a re-design from the ground up of how to handle |
47 |
repositories and ebuilds within them, with efficiency and speed in mind. |
48 |
|
49 |
Unfortunately, pkgcore never gained enough developer support to to keep |
50 |
up with EAPI changes and the miriad of portage options that kept |
51 |
expanding. So pkgcore never ended up replacing portage as our main |
52 |
package manager. This is one reason why Brian eventually gave up on it |
53 |
and subsequently left gentoo. |
54 |
|
55 |
|
56 |
While I put a lot of effort into the re-write of repoman so that the |
57 |
code is maintainable, more easily extendable. It will never be as fast |
58 |
as pkgcheck due to the legacy portage code it is based on. |
59 |
|
60 |
Yes, repoman could be updated to do more checks and catch everything |
61 |
that pkgcheck does. It can be made to do many of the checks in |
62 |
parrallel, but it will never be as efficient as pkgcheck due to the |
63 |
underlying portage code it uses. I largely stopped working on |
64 |
repoman/gentoo due to the near constant bitching about nearly everything |
65 |
I did from another (ahem) gentoo developer which shall remain un-named. |
66 |
|
67 |
|
68 |
My only concern about the possible deprecation of repoman, is the fact |
69 |
that is is based on the current official and primary package manager |
70 |
code used in gentoo. While pkgcheck is not. For that reason, |
71 |
changes made to the core code is mostly automatically carried |
72 |
over to repoman without need to change repoman as well. Ideally, |
73 |
development should be moved to pkgcore and portage be put into |
74 |
maintenance mode only. With pkgcore taking over as the primary package |
75 |
manager for gentoo. But Zac is just too damn efficient at |
76 |
updating/fixing portage bugs! As well as adding features and |
77 |
options... (too damn many to remember half of them now) |