| 1 |
On 10/28/2015 12:23 PM, Anthony G. Basile wrote: |
| 2 |
> |
| 3 |
> A properly designed sub-USE flag would be useful here and clearly better |
| 4 |
> than our REQUIRED_USE. I think REQUIRED_USE is fine for heterogeneous |
| 5 |
> cases, but not when you have something like curl where you can either |
| 6 |
> turn ssl on or off. If it is off, nothing more needs to be specified, |
| 7 |
> if it is on, then you must further specify one and exactly one ssl |
| 8 |
> provider. |
| 9 |
> |
| 10 |
|
| 11 |
Uhm, curl makes use of REQUIRED_USE heavily, otherwise the use_expand |
| 12 |
would not work: |
| 13 |
> REQUIRED_USE=" |
| 14 |
> curl_ssl_winssl? ( elibc_Winnt ) |
| 15 |
> threads? ( !adns ) |
| 16 |
> ssl? ( |
| 17 |
> ^^ ( |
| 18 |
> curl_ssl_axtls |
| 19 |
> curl_ssl_gnutls |
| 20 |
> curl_ssl_libressl |
| 21 |
> curl_ssl_openssl |
| 22 |
> curl_ssl_nss |
| 23 |
> curl_ssl_polarssl |
| 24 |
> curl_ssl_winssl |
| 25 |
> ) |
| 26 |
> )" |
| 27 |
|
| 28 |
With the providers syntax from exherbo, this monster would basically be |
| 29 |
gone. |
Archives