W dniu śro, 10.01.2018 o godzinie 21∶45 -0500, użytkownik Mike Gilbert
napisał:
> On Wed, Jan 10, 2018 at 5:56 PM, Zac Medico <zmed...@gentoo.org> wrote:
> > On 01/10/2018 02:24 PM, Michał Górny wrote:
> > > Enable repoman checks on exp profiles by default to improve
> > > the dependency graph integrity on those profiles and help them on their
> > > way towards stable status. This is possible now that the dependency
> > > graph problems are warnings rather than errors.
> > > ---
> > >  repoman/pym/repoman/argparser.py | 2 +-
> > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > > 
> > > diff --git a/repoman/pym/repoman/argparser.py 
> > > b/repoman/pym/repoman/argparser.py
> > > index f32972288..d49147366 100644
> > > --- a/repoman/pym/repoman/argparser.py
> > > +++ b/repoman/pym/repoman/argparser.py
> > > @@ -167,7 +167,7 @@ def parse_args(argv, qahelp, repoman_default_opts):
> > > 
> > >       parser.add_argument(
> > >               '-e', '--include-exp-profiles', choices=('y', 'n'), 
> > > metavar='<y|n>',
> > > -             default=False,
> > > +             default='y',
> > >               help='include exp profiles in dependency checks')
> > > 
> > >       parser.add_argument(
> > > 
> > 
> > We have dev and exp profiles disabled by default because the time
> > consumed by repoman is proportional to the number of profiles.
> > 
> > The current counts are as follows:
> > 
> > stable 87
> > dev 88
> > exp 149
> 
> Yeah, I really don't like the idea of making repoman even slower by default.

The alternative is to go for all profiles being stable. Because
accepting developers randomly breaking profiles 'because repoman speed'
is not acceptable.

> 
> Also, I'm not sure why exp profiles are to be treated as more stable
> than dev profiles. I was always under the impression that the reverse
> was true, though I cannot recall where that assumption was formed.

A lot of people have different impressions, and noone really ever
bothered trying to document what means what. 'exp' is easier to change
because the argument takes y/n, so '-e n' can easily be done by user,
unlike '-d' that is pure boolean flag.

> 
> I would like to see this formally documented somewhere before we start
> changing repoman defaults.

It's part of my proposal for Sunday's Council meeting.

-- 
Best regards,
Michał Górny


Reply via email to