| 1 |
On Sat, Dec 4, 2010 at 06:02, Ryan Hill <dirtyepic@g.o> wrote: |
| 2 |
> On Sat, 04 Dec 2010 03:29:45 +0100 |
| 3 |
> Diego Elio Pettenò <flameeyes@×××××.com> wrote: |
| 4 |
> |
| 5 |
>> Il giorno ven, 03/12/2010 alle 19.46 -0600, Ryan Hill ha scritto: |
| 6 |
>> > |
| 7 |
>> > This has come up enough times that we should write some common code. |
| 8 |
>> |
| 9 |
>> Or resume the idea to simply provide a separate variable for |
| 10 |
>> number-of-jobs rather than relying purely on MAKEOPTS. |
| 11 |
> |
| 12 |
> That's not a bad idea, but I think we'd still need to fall back to MAKEOPTS |
| 13 |
> parsing if that variable was not set. |
| 14 |
|
| 15 |
This has been discussed here two years ago, to the day. But |
| 16 |
unfortunately the discussion didn't when to a decision [1]. |
| 17 |
I filled a QA tracker bug [2] to summarize the proposed solutions, and |
| 18 |
re-open the discussion. |
| 19 |
In my opinion, just filtering out the --load-average option, and |
| 20 |
keeping the --jobs value is *bad*, as I putted it with serious reasons |
| 21 |
(detailed in the tracker). So I wonder if I can have a setting saying |
| 22 |
"if that build system cannot adapt to the current load, then I want a |
| 23 |
--jobs value of 1, or 2, but not 4"? |
| 24 |
As having SCONSOPTS WAFOPTS ANTOPTS CMAKEOPTS and so on variables is a |
| 25 |
ugly solution, and would require a lot of eclasses to be changed, I |
| 26 |
chose the fallback of having a empty MAKEOPTS, and to rely on |
| 27 |
portage's --jobs and --load-average options. |
| 28 |
|
| 29 |
1 http://archives.gentoo.org/gentoo-dev/msg_750e33f68b16d971dff1f40dd9145e56.xml |
| 30 |
2 https://bugs.gentoo.org/337831 |
| 31 |
|
| 32 |
-- |
| 33 |
Cyprien Nicolas |
| 34 |
Gentoo Lisp Project contributor |
| 35 |
Fulax on #gentoo-lisp |
Archives