| 1 |
I noticed there is some fairly involved logic in the GCC 3.1 ebuild |
| 2 |
that attempts to deduce whether or not GCC 2.95.3 is already installed |
| 3 |
on the system, in which case it tries to play nice and install itself |
| 4 |
alongside it without trampling it. This results in 2.95.3 remaining |
| 5 |
as "gcc", while 3.1 becomes "gcc-3.1". This is really helpful, but I |
| 6 |
was wondering if it was an officially supported state of affairs. |
| 7 |
|
| 8 |
If so, it would be great to simply add a line such as CC="gcc-3.1" to |
| 9 |
/etc/make.conf and have all ebuilds use GCC 3.1. The CC definition |
| 10 |
could be commented out for pesky problematic software. Most |
| 11 |
autoconf-based packages will honor this automatically, but not all |
| 12 |
ebuilds are autoconf-based. |
| 13 |
|
| 14 |
I don't want to file lots of petty bug reports about triviata, but |
| 15 |
should failure to honor the CC environment variable be considered a |
| 16 |
bug? The first hurdle I have hit is portage itself. |
| 17 |
|
| 18 |
-- |
| 19 |
Robert Coie <rac@××××××××××.jp> |
| 20 |
Implementor, Apropos Ltd. |
Archives