1 |
Am Mittwoch, 6. Dezember 2017, 08:00:40 CET schrieb Raffaele Belardi: |
2 |
> One (~x86) LXDE system completed the switch with no problem, the other |
3 |
> (~amd64) built all except two packaged (sdlmame and torcs) which did not |
4 |
> build with gcc-7.2 even before the switch to 17.0. |
5 |
> |
6 |
> Gentoo devs and arch testers did a good job as usual. |
7 |
> |
8 |
> I'll do the switch on the Gnome system in the next days but up to now I can |
9 |
> say that the switch to 17.0 is a _lot_ less painful than switching major |
10 |
> compiler version. |
11 |
> |
12 |
> raffaele |
13 |
|
14 |
I'll add my support for this, the migration was almost completely painless on |
15 |
all three of my systems. There were only a few exceptions: |
16 |
|
17 |
- I hit the aforementioned cdrdao failure (which I opted to solve via USE |
18 |
flag). |
19 |
|
20 |
- On my laptop I hit the ICU incompatibility with qtwebengine because I had |
21 |
firefox install with USE="system-icu" (which was masked at some point in the |
22 |
13.0 profiles, which is why I didn't hit this until now), so I unset the USE |
23 |
flag again, but that caused some post-migration emerging, which was annoying, |
24 |
but not horrible. |
25 |
|
26 |
- On my desktop rust failed for some reason (almost at the end, of course, |
27 |
when building rustdoc), but simply emerging it again afterwards worked. |
28 |
|
29 |
- On my desktop I also hit the mupen64plus-ui-console build failure. |
30 |
|
31 |
Note that all of these have open bugs (except for the rust one, but I expect |
32 |
that simply came from an inconsistency due to the unfinished "emerge -e |
33 |
@world", or a random bitflip, or something similarly ephemeral). |
34 |
|
35 |
(Oh, and libsidplay failed because its maintainer is apparently too busy to |
36 |
fix its incompatibility with app-shells/dash, but that had nothing to do with |
37 |
the migration per se.) |
38 |
|
39 |
I'll also add -- not because I was worried about it but because the |
40 |
possibility was mentioned in another thread -- that I can't say I'm noticing |
41 |
any performance hits, even on my 11 year old Athlon64 X2. (IIUC, per the GCC |
42 |
manual, PIC and PIE only affect startup time, not runtime, so this result |
43 |
should be expected.) |
44 |
|
45 |
-- |
46 |
Marc Joliet |
47 |
-- |
48 |
"People who think they know everything really annoy those of us who know we |
49 |
don't" - Bjarne Stroustrup |