1 |
On 2016-01-24 18:44, Michael Orlitzky wrote: |
2 |
> On 01/24/2016 06:29 PM, Aaron W. Swenson wrote: |
3 |
> > Okay, provided that the new USE_EXPAND is okay for POSTGRES_TARGETS, |
4 |
> > attached are the eclasses that I'll commit to the tree. |
5 |
> > |
6 |
> |
7 |
> > case ${EAPI:-0} in |
8 |
> > 0|1|2|3|4) die "postgres-multi.eclass requires EAPI 5 or higher" ;; |
9 |
> > *) ;; |
10 |
> > esac |
11 |
> |
12 |
> Does this really work with EAPI=6? I didn't try, but it looks like it |
13 |
> would need an eapply_user somewhere in src_prepare. And, pedantry |
14 |
> warning, there's no such thing as "EAPI 5 or higher." The lawyers will |
15 |
> tell you to do something like this instead (stolen from git-r3): |
16 |
> |
17 |
> case "${EAPI:-0}" in |
18 |
> 5|6) |
19 |
> ;; |
20 |
> *) |
21 |
> die "Unsupported EAPI=${EAPI} (unknown) for ${ECLASS}" |
22 |
> ;; |
23 |
> esac |
24 |
> |
25 |
> That will require an edit for every new EAPI, but prevents weird crashes |
26 |
> from things like a missing eapply_user. |
27 |
> |
28 |
> |
29 |
|
30 |
Thank you. |
31 |
|
32 |
I've added the eapply_user to postgres-multi and modified the case |
33 |
condition to match the hot goods you're selling on the sly. |