1 |
Ciaran McCreesh wrote: |
2 |
> On Thu, 20 Mar 2008 18:02:31 -0600 |
3 |
> Ryan Hill <dirtyepic@g.o> wrote: |
4 |
>>> How about just checking EUID == 0 in src_test and skip the tests |
5 |
>>> (with a ewarn message) if it doesn't match your needs? |
6 |
>> I thought I remembered someone raising a stink about checking |
7 |
>> permissions being a race condition because the condition can change |
8 |
>> between the checking and performing the action. Maybe that was only |
9 |
>> about write permissions... |
10 |
> |
11 |
> That's in cases like if [[ -w foo ]] ; then blah foo ; fi . Checking a |
12 |
> variable that's constant for the phase is not racy. |
13 |
> |
14 |
>> If checking EUID is an acceptable method, |
15 |
>> it should be relatively simple to write an eclass to handle the |
16 |
>> common cases and have ebuilds use that instead of checking $FEATURES |
17 |
>> (which i do think is a bug). |
18 |
> |
19 |
> Does it really need an eclass for something so simple? |
20 |
|
21 |
Considering the number of ebuilds that check FEATURES, disable tests altogether, |
22 |
or just do nothing and fail for half our users, against the number that do it |
23 |
correctly... apparently yeah. The simpler we make it to use, the more people |
24 |
will use it. |
25 |
|
26 |
On the other hand, it might be better to just document the common cases along |
27 |
with some example code in the Dev Guide. I'd be happy with doing that. |
28 |
|
29 |
|
30 |
-- |
31 |
fonts, gcc-porting, by design, by neglect |
32 |
mips, treecleaner, for a fact or just for effect |
33 |
wxwidgets @ gentoo EFFD 380E 047A 4B51 D2BD C64F 8AA8 8346 F9A4 0662 |