| 1 |
On 12/08/2015 12:59, covici@××××××××××.com wrote: |
| 2 |
> Holger Hoffstätte wrote: |
| 3 |
> |
| 4 |
>> On Wed, 12 Aug 2015 05:18:54 -0400, covici wrote: |
| 5 |
>> |
| 6 |
>>> Holger Hoffstätte wrote: |
| 7 |
>>> |
| 8 |
>>>> On Tue, 11 Aug 2015 12:01:50 +0200, Helmut Jarausch wrote: |
| 9 |
>>>> |
| 10 |
>>>>> On 08/10/2015 05:29:04 PM, James wrote: |
| 11 |
>>>>>> Hello, |
| 12 |
>>>>>> |
| 13 |
>>>>>> Just for grins, I'm thinking about installing a gentoo system that |
| 14 |
>>>>>> is (~) testing and newer codes and setting gcc-.5.2 as the default |
| 15 |
>>>>>> compiler. Anyone doing this yet? Will it mostly work or should I |
| 16 |
>>>>>> wait? Workstation with lxqt-0.9x. |
| 17 |
>>>>> |
| 18 |
>>>>> I don't use lxqt here, but ... |
| 19 |
>>>>> I'm using gcc-5.2.0 very successfully here. I am not aware of a |
| 20 |
>>>>> single package which doesn't compile with this version. |
| 21 |
>>>>> And the kernel itself compiles just fine (currentlich running |
| 22 |
>>>>> 4.2.0-rc6 compiled with gcc-5.2.0) |
| 23 |
>>>> |
| 24 |
>>>> +1 |
| 25 |
>>>> |
| 26 |
>>>> I just updated all my ~amd64 multilib systems to 5.2 and everything |
| 27 |
>>>> still works without a hitch, including acroread & skype. 4.9.3 is gone. |
| 28 |
>>>> :) |
| 29 |
>>>> |
| 30 |
>>>>> One caveat though when upgrading a system which has been built with |
| 31 |
>>>>> gcc-x for x < 5. |
| 32 |
>>>>> The (pure) C++ libraries aren't compatible with older versions so |
| 33 |
>>>>> that all packages which link to libstdc++ or libsupc++ have to be |
| 34 |
>>>>> rebuilt. Packages which just use C aren't effected at all. |
| 35 |
>>>> |
| 36 |
>>>> Interestingly I didn't have to rebuild anything (yet), which I find |
| 37 |
>>>> surprising given the changes in the ABI and std::string. I expected at |
| 38 |
>>>> least a few problems, |
| 39 |
>>>> but so far - nothing. |
| 40 |
>>> |
| 41 |
>>> I can't even emerge gcc-5.2.0 because of missing keyword, as of a sync |
| 42 |
>>> of yesterday, so what is happening with that? |
| 43 |
>> |
| 44 |
>> The ebuild currently doesn't have any keywords, so enable them: |
| 45 |
>> |
| 46 |
>> $grep gcc /etc/portage/profile/package.accept_keywords |
| 47 |
>>> =sys-devel/gcc-5.2.0 ** (or your specific arch instead of **) |
| 48 |
> |
| 49 |
> But why do they have no keywords? Does this mean the package is not |
| 50 |
> ready for prime time or much else? |
| 51 |
> |
| 52 |
|
| 53 |
|
| 54 |
It's normal for new bleeding-edge versions to be in the tree without |
| 55 |
keywords. It means the code compiled and seems to do something useful |
| 56 |
for at least one person (the dev) but cannot be guaranteed to work yet |
| 57 |
on any arch. Hence no keywords = will not install on any arch without |
| 58 |
user intervention |
| 59 |
|
| 60 |
|
| 61 |
|
| 62 |
-- |
| 63 |
Alan McKinnon |
| 64 |
alan.mckinnon@×××××.com |
Archives