1 |
On Wed, 17 Dec 2014 23:59:59 +0200 |
2 |
Alan McKinnon <alan.mckinnon@×××××.com> wrote: |
3 |
|
4 |
> On 18/12/2014 04:45, Harry Putnam wrote: |
5 |
> > Is there any advantage one way or the other emerging firefox.bin vs |
6 |
> > firefox? |
7 |
> |
8 |
> Depends on your needs: |
9 |
> |
10 |
> firefox: |
11 |
> - pro: you get all the USE flags |
12 |
> - pro: you don't get bundled libs from Mozilla, the ebuild can use |
13 |
> system libs |
14 |
> - pro: the compiled binaries are integrated into gentoo like other |
15 |
> ebuilds |
16 |
> - con: slow compiles. I have 8 i7 cores and 16G. the merge takes 20-35 |
17 |
> minutes... |
18 |
> |
19 |
> firefox-bin: |
20 |
> - pro: fast install. It's a binary package |
21 |
> - con: you get all of Mozilla's bundled libs |
22 |
> - con: No USE, no choices. If Mozilla eg decides to ship with |
23 |
> pulseaudio, then that is what you must have on your end |
24 |
> - con: poor integration with the rest of your system. Files go where |
25 |
> Mozilla says they go, the devs can only do so much to make stuff |
26 |
> standard. |
27 |
|
28 |
Those are good lists. The only thing I can think to add is that |
29 |
firefox-bin is built with "Profile Guided Optimization"; the firefox |
30 |
package has the pgo USE flag for that, but it's forced off because it |
31 |
doesn't work and upstream doesn't support it. |
32 |
|
33 |
Building with PGO roughly doubles compile time, as firefox has to be |
34 |
built twice. I don't know what optimization gains there are. |
35 |
|
36 |
> As I see it, go with firefox unless you can't spend the cpu cycles to |
37 |
> build it locally. That's true of almost all -bin packages |
38 |
|
39 |
+1 |