Re: [gentoo-dev] Re: [RFC] Allow bash-4.0 features in EAPI="3" ebuilds

[Patrick Lauer]

On Sunday 24 May 2009 23:22:21 Ciaran McCreesh wrote:
> On Sun, 24 May 2009 23:16:13 +0200
>
> Patrick Lauer <patrick@g.o> wrote:
> > Okay, yes, Mr. Long was quite rude there (trying to fight fire with
> > fire I guess). But in this case you're discussing rather subjective
> > things again (how often is it the case that you don't have a cache?)
> > that might not even be a problem.
>
> This is not in the least bit subjective. You don't have cache:
>
> * for any overlays you use
Only partially correct (but you knew that already, so I won't bother repeating 
it)
And with overlays you have _no_ guarantees anyway. Plus portage spews a nice 
warning if you play around with eclasses, which many users parse as an error. 
So that's not an issue anyway ... overlays are unsupported territory where the 
only assumption you can make is that things might not work (but surprisingly 
often they do)

> * often enough for the main tree that we get people asking about it in
>   #paludis, since Paludis warns if it encounters stale cache files.
Haven't seen that with portage (well duh), and if it really is a problem maybe 
we should ask grobian how he made the prefix rsync checkout consistent. 
Y'know, if it is a problem fix the problem.

> We know full well that this is a real problem. Stop pretending that it
> isn't.
Well, we don't. Hmm, who is we in this context? Would be much nicer if we 
stopped using the pluralis majestatis to make us look more important. Anyways, 
if it is a problem it's mostly an issue of the cache generation, which is 
trivially fixed, or it's not a problem, in which case it is fixed already. Or 
it is an issue for everyone doing unsupported things, in which case ... it is 
not an issue. Because it's unsupported. 

So what was our point again? 

Anyways, this is going round and round in circles until someone gets dizzy and 
throws up. Not the best way to "discuss", and I'm getting really bored with 
it.