1 |
On Tue, 13 Sep 2011 21:02:28 -0500 |
2 |
Donnie Berkholz <dberkholz@g.o> wrote: |
3 |
> On 17:56 Tue 13 Sep , Mike Frysinger wrote: |
4 |
> > useful enough for EAPI ? or should i just stick it into |
5 |
> > eutils.eclass ? OR BOTH !? |
6 |
> |
7 |
> I prefer to avoid EAPI whenever possible, as it just makes things |
8 |
> slower and more complex. |
9 |
|
10 |
Sticking it in an EAPI *shouldn't* be slow and more complex. There are |
11 |
three reasons why it is, and they should all be within Gentoo's |
12 |
ability to solve. |
13 |
|
14 |
The first reason is that when we did what was then called EAPI 3, |
15 |
several Council members refused to put in more than one hour's work |
16 |
every month. To get an EAPI out quickly, we need Council members who |
17 |
are prepared to do a bit of homework, and to read proposals before a |
18 |
meeting and to comment on mailing lists rather than only bringing up |
19 |
questions (most of which have already been answered on the lists) at |
20 |
meetings. That shouldn't be too much to ask, and if it is, Council |
21 |
members should be prepared to delegate. |
22 |
|
23 |
The second is that it's impossible to get an accurate estimate from the |
24 |
Portage people for how long it will take to implement something. We |
25 |
were assured before the then-EAPI-3 proposals were submitted to the |
26 |
Council that all would be easy and quick to implement in Portage. We |
27 |
were told after approval that implementation would take a month, when |
28 |
it took a year to get just partial implementations of some features. |
29 |
This needs to be addressed -- to a certain extent we can drop features, |
30 |
but EAPI 4 currently has nasty problems (prefix flag needs to be in |
31 |
IUSE, and people are refusing to do that) due to one of two |
32 |
interdependent features being dropped (strict IUSE) without the |
33 |
standard wording for the other (use dependency defaults) being changed. |
34 |
|
35 |
The third is that there are a few people intent on preventing any new |
36 |
EAPI from ever happening. The solution here is to tell them that the |
37 |
decision has been made, that Gentoo is going to use EAPIs and a |
38 |
specification whether they like it or not, and that they can either help |
39 |
or keep quiet. No-one has the time to deal with a small group of |
40 |
individuals who pop up and yell "PMS sucks! EAPIs are bad! Portage is |
41 |
reality! Code to an implementation not a standard!" every time anyone |
42 |
asks for a new feature. |
43 |
|
44 |
None of this should be difficult. |
45 |
|
46 |
-- |
47 |
Ciaran McCreesh |