1 |
On Sunday, 5. October 2008, Ulrich Mueller wrote: |
2 |
> >>>>> On Sun, 5 Oct 2008, Robert Buchholz wrote: |
3 |
> >> > |
4 |
> >> > It's not. If you want to have default DOCS then you should loop |
5 |
> >> > through the items and check with [[ -e ]] before trying to |
6 |
> >> > install them. |
7 |
> >> |
8 |
> >> So, maybe just do a 'dodoc "${DOCS}"' and omit the die? Then it won't |
9 |
> >> fail but the warning message would be preserved. |
10 |
> > |
11 |
> > I understood Petteri's comment to be related to the default case |
12 |
> > (i.e. the else-branch), and I have to agree there: Ebuilds that do |
13 |
> > not override src_install should not emit a warning when some |
14 |
> > ChangeLog file is missing that the ebuild never specified to |
15 |
> > install. |
16 |
> |
17 |
> The default would be an empty DOCS variable, or did I miss something? |
18 |
|
19 |
Correct. |
20 |
|
21 |
> So if the ebuild includes non-existing files in DOCS, then why would |
22 |
> you want to suppress the warnings? |
23 |
|
24 |
I don't. My point was that the default action on an empty DOCS variable is |
25 |
to "dodoc AUTHORS ChangeLog NEWS README", and this should not emit |
26 |
warnings, because it is merely a heuristic. |
27 |
|
28 |
To be clearer: |
29 |
else |
30 |
- # No die here because we don't know if any of these exist |
31 |
- dodoc AUTHORS ChangeLog NEWS README |
32 |
+ for x in AUTHORS ChangeLog NEWS README; do |
33 |
+ if [ -e ${x} ]; then |
34 |
+ dodoc ${x} || die "dodoc ${x} failed" |
35 |
+ fi |
36 |
+ done |
37 |
fi |
38 |
|
39 |
|
40 |
Robert |