1 |
Brian Harring wrote: |
2 |
|
3 |
>> And no, I don't think that Gentoo should fully support reduced-@system |
4 |
>> builds, but there is no harm in making them more of a viable option. |
5 |
> |
6 |
> Personally... I think gentoo should aim for it actually. Question is |
7 |
> how close we can get to it w/out overly burdening developers. |
8 |
> |
9 |
I always thought of @system as providing a POSIX-compatible userland, |
10 |
exactly so that deps wouldn't need to be listed. eg every POSIX compatible |
11 |
system has to have find, sed etc[1] and of course ed which is missing, |
12 |
annoyingly enough for scripting. |
13 |
|
14 |
With regard to having a slimmed-down @system it might appear to make sense |
15 |
to look at moving to only allowing POSIX options, testing perhaps with |
16 |
sys-apps/minised[2] (assuming we'd have a virtual in @system) which I |
17 |
haven't tried. AIUI however there's far too many sed GNUisms to go down |
18 |
that road. |
19 |
|
20 |
Even on FreeBSD they're still using gsed; the fact that people will not stop |
21 |
using GNU-style regexes-- and that they need it for their own base system-- |
22 |
is only driving them to work on including the same extensions in their own |
23 |
software.[3] |
24 |
|
25 |
Given that we're not about to clear the GNUisms from the tree, it's been "a |
26 |
system package since at least 2004" with dependencies on versions "as high |
27 |
as 4.0.5, which went stable in 2003" I'd concur that the existing ebuild/ |
28 |
eclass dependencies should be trimmed as and when, if it can't be |
29 |
automated. |
30 |
|
31 |
Hopefully at some point there'll be another implementation with the same or |
32 |
similar capabilities, which might even make it into POSIX. Until then it |
33 |
would appear Gentoo needs gsed just like it needs bash, afaict. An embedded |
34 |
system might well have a different profile; presumably applications would |
35 |
not be built on the target tho. |
36 |
|
37 |
[1] http://pubs.opengroup.org/onlinepubs/9699919799/utilities/contents.html |
38 |
[2] http://www.exactcode.de/oss/minised/ |
39 |
[3] See point 3 at: |
40 |
http://freebsd.1045724.n5.nabble.com/RFC-Replacing-our-regex-implementation-td4380832.html |
41 |
-- |
42 |
#friendly-coders -- We're friendly but we're not /that/ friendly ;-) |