| 1 |
>>>>> On Mon, 16 Feb 2015, David Leverton wrote: |
| 2 |
|
| 3 |
> The original implementation for --disable-dependency-tracking in |
| 4 |
> Portage really did check without the "--", but it was changed in |
| 5 |
> 7d7387b (2012-08-29) at the same time that --disable-silent-rules |
| 6 |
> was added. Paludis follows the spec as it currently is and doesn't |
| 7 |
> include the "--". As this is such a small detail I'm not going to |
| 8 |
> complain too much, but technically it is a retroactive change.... |
| 9 |
|
| 10 |
I don't remember why we don't include the "--", but the earlist |
| 11 |
mention of it seems to be this: |
| 12 |
https://bugs.gentoo.org/show_bug.cgi?id=211529#c24 |
| 13 |
(I suspect that it was omitted simply because grepping for a string |
| 14 |
starting with a hyphen is slightly more complicated ...) |
| 15 |
|
| 16 |
Anyway, Portage includes the "--" since some time and there seem to be |
| 17 |
no issues with it. More importantly, EAPI 6 will have --docdir and |
| 18 |
--htmldir, and I fear that matching them without the "--" might lead |
| 19 |
to false positives. |
| 20 |
|
| 21 |
So our choices are a) inconsistent behaviour for EAPI 6, matching |
| 22 |
"disable-silent-rules" but "--docdir", b) a complicated spec with |
| 23 |
"disable-silent-rules" in EAPI 5 (and not matching Portage behaviour) |
| 24 |
but "--disable-silent-rules" in EAPI 6, or c) a small retroactive |
| 25 |
change. To me c) looks like the lesser evil. |
| 26 |
|
| 27 |
Ulrich |
Archives