| 1 |
>>>>> On Wed, 22 Jun 2016, Daniel Campbell wrote: |
| 2 |
|
| 3 |
> Sounds decent to me. Will any eclasses be affected? |
| 4 |
|
| 5 |
Certainly l10n.eclass will need some tweaking. |
| 6 |
|
| 7 |
> What sort of changes will need to be made in ebuilds? I ask so I |
| 8 |
> have a list of packages I can draw up that need this changed so I |
| 9 |
> can get things done when everything's in place with the guide. |
| 10 |
|
| 11 |
For the simple two- and three-letter language codes it will be only |
| 12 |
renaming from linguas_* to l10n_*. Composite codes like pt_BR have the |
| 13 |
underscore replaced by a hyphen (i.e., pt-BR now). There are very few |
| 14 |
cases containing a script or variant subtag: |
| 15 |
|
| 16 |
ca-valencia - Catalan (Valencian) |
| 17 |
kmr-Latn - Northern Kurdish (Latin script) |
| 18 |
sr-Latn - Serbian (Latin script) |
| 19 |
|
| 20 |
Presumably these will require some special conditions (in most ebuilds |
| 21 |
they are already special now with their old LINGUAS syntax). |
| 22 |
|
| 23 |
I'll file a tracker bug later. Currently there are 170 packages having |
| 24 |
linguas_* use conditionals in SRC_URI or *DEPEND. As soon as these are |
| 25 |
converted to use l10n_* instead, we can do the rest in one atomic |
| 26 |
commit, along with the removal of LINGUAS from USE_EXPAND. |
| 27 |
|
| 28 |
Ulrich |
Archives