| 1 |
On 17:02 Thu 22 Jan , Ciaran McCreesh wrote: |
| 2 |
> On Thu, 22 Jan 2009 08:56:23 -0800 |
| 3 |
> Donnie Berkholz <dberkholz@g.o> wrote: |
| 4 |
> > Can this be fixed by adding bash dependencies to things using new |
| 5 |
> > features? As long as we keep them out of the build path of bash, |
| 6 |
> > things ought to work. |
| 7 |
> |
| 8 |
> Only if you're guaranteed bash 3.1 on boxes that do metadata |
| 9 |
> generation. Which means it won't work for overlays. |
| 10 |
|
| 11 |
I'm not an expert on metadata generation, so please tell me if I'm wrong |
| 12 |
here. Most if not all overlays don't ship pregenerated metadata, which |
| 13 |
means users have to generate it locally. Without doing so, they cannot |
| 14 |
install anything from that overlay that uses bash features they lack. |
| 15 |
They can still install things from that overlay that don't use the new |
| 16 |
bash features. |
| 17 |
|
| 18 |
Consequently, packages in overlays using new bash features won't work |
| 19 |
till you upgrade your bash. They also won't be able to give you a good |
| 20 |
error about how to fix it, because you can't guarantee that you'll be |
| 21 |
able to parse the ebuild/eclass as far as the bash dependency. |
| 22 |
|
| 23 |
I guess a GLEP 42 news item would sort of work, but I really wish we had |
| 24 |
tree dependencies. Another option would be to make overlay maintainers |
| 25 |
generate their own metadata. |
| 26 |
|
| 27 |
> > Then we could add a repoman check for new features, if we wanted. |
| 28 |
> |
| 29 |
> Can't do that unless you write a feature-complete bash parser and |
| 30 |
> pretend-execute every possible path an ebuild can take... |
| 31 |
|
| 32 |
We're getting into pretty weird territory here -- if you had slotted |
| 33 |
bash versions, you could do bash-3.0 -n foo.ebuild, bash-3.1 -n |
| 34 |
foo.ebuild, etc. |
| 35 |
|
| 36 |
-- |
| 37 |
Thanks, |
| 38 |
Donnie |
| 39 |
|
| 40 |
Donnie Berkholz |
| 41 |
Developer, Gentoo Linux |
| 42 |
Blog: http://dberkholz.wordpress.com |
Archives