| 1 |
On Wednesday 18 July 2012 13:29:41 Ciaran McCreesh wrote: |
| 2 |
> On Wed, 18 Jul 2012 18:18:35 +0200 "Andreas K. Huettel" wrote: |
| 3 |
> > > On Wed, Jul 18, 2012 at 5:33 PM, hasufell <hasufell@g.o> |
| 4 |
> > > |
| 5 |
> > > wrote: |
| 6 |
> > > > "epatch" is so widely used and basic that I wonder why it's still |
| 7 |
> > > > not implemented as a real helper function. |
| 8 |
> > > |
| 9 |
> > > Because then its harder to change, it must be in PMS, otherwise you |
| 10 |
> > > have to do things like test which version of epatch the package |
| 11 |
> > > manager provides....sounds a lot like EAPI :) |
| 12 |
> > |
| 13 |
> > You know, that's actually a pretty good case *for* base.eclass, |
| 14 |
> > eutils.eclass and similar... we should probably move more functions |
| 15 |
> > there... :D |
| 16 |
> |
| 17 |
> I'm not sure that having to make sure you don't break ten thousand |
| 18 |
> packages whenever you make a change is a good case... When it's EAPI |
| 19 |
> controlled, if a change causes problems, it doesn't break anything. |
| 20 |
|
| 21 |
and the obvious con is that it's hard to add new features and extend |
| 22 |
implementation details without also upgrading all EAPI aspects. locking down |
| 23 |
EAPI is great for the format of the file and for simpler commands (like most of |
| 24 |
the install funcs), but for more complicated functions, an eclass is nicer. |
| 25 |
-mike |
Archives