| 1 |
On Sun, 03 Mar 2013 23:05:28 +0100 |
| 2 |
hasufell <hasufell@g.o> wrote: |
| 3 |
|
| 4 |
> -----BEGIN PGP SIGNED MESSAGE----- |
| 5 |
> Hash: SHA1 |
| 6 |
> |
| 7 |
> On 03/03/2013 06:24 PM, Ryan Hill wrote: |
| 8 |
> > On Sun, 03 Mar 2013 15:42:56 +0100 hasufell <hasufell@g.o> |
| 9 |
> > wrote: |
| 10 |
> > |
| 11 |
> >> What do we have useflags for in gentoo? |
| 12 |
> > |
| 13 |
> > Not for conditional patching, that's for sure. |
| 14 |
> > |
| 15 |
> |
| 16 |
> That simply diverges from reality. |
| 17 |
> |
| 18 |
> # qgrep epatch | grep 'use\ ' | wc -l |
| 19 |
> 476 |
| 20 |
> |
| 21 |
> Since conditional seds are practically the same... |
| 22 |
> |
| 23 |
> # qgrep ' sed ' | grep 'use\ ' | wc -l |
| 24 |
> 447 |
| 25 |
> |
| 26 |
> And those are just lazy greps not catching if else foo syntax. |
| 27 |
|
| 28 |
Sometimes it's unavoidable. Sometimes people don't think before they do |
| 29 |
something. If the patch makes a fundamental change in the package that can't |
| 30 |
be controlled another way (say --configure flags or defines) then, yes, you may |
| 31 |
have to use conditional patching. I'm thinking of something like infinality or |
| 32 |
x32 support. In both those cases we're adding a feature, not making a bug |
| 33 |
fix. That's an important distinction. USE flags exist to give the user a |
| 34 |
choice between A and B. If choice A is "package doesn't build" then there's |
| 35 |
really no choice at all. You've just added a new way for a package to fail. We |
| 36 |
already have plenty of those. Please don't add more. |
| 37 |
|
| 38 |
|
| 39 |
-- |
| 40 |
gcc-porting |
| 41 |
toolchain, wxwidgets learn a language baby, it's that kind of place |
| 42 |
@ gentoo.org where low card is hunger and high card is taste |
Archives