Re: [gentoo-dev] Removing <herd>no-herd</herd>?

[Michał Górny]

On Sat, 10 Sep 2011 11:22:16 +0300
Markos Chandras <hwoarang@g.o> wrote:

> On 10/09/2011 11:15 πμ, Mike Gilbert wrote:
> > On Sat, Sep 10, 2011 at 4:04 AM, Markos Chandras
> > <hwoarang@g.o> wrote:
> >> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA512
> >>
> >> On 10/09/2011 11:01 πμ, Michał Górny wrote:
> >>> Hello,
> >>>
> >>> Is there a real reason to have <herd>no-herd</herd>? As I see
> >>> it, it's just an ugly hack which all programs have to learn and
> >>> hack for no benefit.
> >>>
> >> What is the problem with that? Why is it an ugly hack?
> >>
> >
> > To a simple program/user, it basically says that the package
> > belongs to a herd called "no-herd", rather than not belonging to
> > any herd.
> Well it is pretty obvious that no-herd means errr no herd.

To a programmer, 'pretty obvious' doesn't work.

> > For example, in IRC, willikins tries to look up the members of
> > "no-herd" when you do !meta -v because the bot lacks a special
> > exception for this odd-ball value.
> Just because willikins behaves like that does not justify the removal
> of this tag. God knows how many utilites and scripts are out there
> using the no-herd tag from metadata.

Please do not mix facts and religion. Facts are that no-herd isn't
really useful. If any utilities actually *rely* on it, they should be
fixed in the first place.

Portage, pkgcore and Paludis have no problems with no <herd/> tag.
The latter two don't even care about 'no-herd'; portage marks it as
a special value and well, that's it.

There's, of course, metadata.dtd too but repoman re-fetches it on
a regular basis so that's no problem.

-- 
Best regards,
Michał Górny