1 |
On Friday 13 April 2007, Ciaran McCreesh wrote: |
2 |
> Mike Frysinger <vapier@g.o> wrote: |
3 |
> > On Friday 13 April 2007, Ciaran McCreesh wrote: |
4 |
> > > These fail: |
5 |
> > > |
6 |
> > > cp somefile dirdoesnotexist/ |
7 |
> > > mv somefile dirdoesnotexist/ |
8 |
> > > ln -s somefile dirdoesnotexist/ |
9 |
> > > dohard somefile dirdoesnotexist/ |
10 |
> > > mkdir dirdoesnotexist/newdir |
11 |
> > > |
12 |
> > > This succeeds: |
13 |
> > > |
14 |
> > > dosym somefile dirdoesnotexist/ |
15 |
> > |
16 |
> > any other obvious statements you care to make ? |
17 |
> > |
18 |
> > compare your standard handwritten Makefile install: target to |
19 |
> > src_install() in an ebuild ... i'll take the much shorter |
20 |
> > src_install() any day |
21 |
> |
22 |
> Except that we already know that it doesn't make a substantial |
23 |
> difference to the lengths of src_installs in the tree. Very few |
24 |
> packages are relying upon the automatic directory creation. |
25 |
|
26 |
i bet you have hard #s to back that up ? since you dont, bloating src_install |
27 |
is not the answer to a marginalized issue |
28 |
|
29 |
> > > The arch teams have been pushing for this for a long time. They're |
30 |
> > > trying to get this enforced, but are having limited success because |
31 |
> > > there's no way for FEATURES=test to become widely used that won't |
32 |
> > > lead to broken user systems. Moving src_test to be always on in |
33 |
> > > future EAPIs is an easy way of helping arch teams achieve this goal |
34 |
> > > without breaking anyone's system in the meantime. |
35 |
> > |
36 |
> > if you have a problem with the QA team then complain to your buddy spb |
37 |
> |
38 |
> Er, no, I'm explaining why enforcing src_test for EAPI 1 will be |
39 |
> helpful for an awful lot of Gentoo developers. |
40 |
|
41 |
except that you back the tree into a corner that it cannot come out of |
42 |
|
43 |
> Please refrain from that kind of comment. It doesn't help anyone. |
44 |
|
45 |
the answer is the same: talk to the QA team to get the tree into a state where |
46 |
having src_test enabled by default is feasible and then the QA team can |
47 |
change the profile |
48 |
|
49 |
enforcing via spec is the wrong way to go here ... spec is for defining how |
50 |
the ebuilds work, not for forcing policy down peoples throats |
51 |
-mike |