1 |
On Wed, Apr 25, 2018 at 08:47:16AM +0200, Ulrich Mueller wrote: |
2 |
> >>>>> On Wed, 25 Apr 2018, Marty E Plummer wrote: |
3 |
> |
4 |
> >> I don't see much code duplication there, so I think it would be |
5 |
> >> cleaner to have a second eclass, rather than adding conditionals to |
6 |
> >> the existing one. |
7 |
> |
8 |
> > I mean if you take into account app-dicts/dictd-* and freedict.eclass; |
9 |
> > without the above pr (which I did originally have with a new eclass, |
10 |
> > until I realized that app-text/dictd's conf file hardcodes /usr/lib/dict |
11 |
> > anyways so the freedict dictionaries get [at least on amd64 and other |
12 |
> > targets which use/will use lib64] installed outside of the search path |
13 |
> > anyways) they are all almost entirely identical with the contents of |
14 |
> > freedict.eclass. |
15 |
> |
16 |
> > Another thought I had was moving the src_install into dict.eclass and |
17 |
> > have freedict inherit it, adding the differences. |
18 |
> |
19 |
> $ egrep -v '^(#|$)' freedict.eclass | wc -l |
20 |
> 18 |
21 |
> |
22 |
> IMHO not worth the effort for 18 lines of code. Keep it simple. |
23 |
> |
24 |
> Ulrich |
25 |
|
26 |
Still, optimization or no, if app-text/dictd is looking at /usr/lib/dict |
27 |
unconditionally and the app-dict/freedict-* dict files are ending up in |
28 |
/usr/lib64/dict conditionally, something needs to be fixed about that. |