1 |
Alan McKinnon schrieb am 04.04.2009 23:55: |
2 |
> On Saturday 04 April 2009 23:42:54 Daniel Pielmeier wrote: |
3 |
>> Joseph schrieb am 04.04.2009 22:48: |
4 |
>>> Is there a way to verify GCC version program was compiled with? |
5 |
>>> I just want to check if all the programs were compiled with latest GCC |
6 |
>>> version as I'm getting an errors at time to time. |
7 |
>> I don't think it is possible to get the compiler or it's version used |
8 |
>> for a specific program. If you are upgrading the compiler it is |
9 |
>> advisable to recompile the complete system so all programs are compiled |
10 |
>> with the same compiler version. Take a look at the gcc upgrading guide |
11 |
>> [1] for the necessary steps you need to follow. |
12 |
>> |
13 |
>> [1] http://www.gentoo.org/doc/en/gcc-upgrading.xml |
14 |
> |
15 |
> This is complete nonsense advice. There is absolutely no need to rebuild the |
16 |
> entire system every time you upgrade compilers, and whoever told you that is |
17 |
> flat out wrong. If the gentoo docs told you that, then they are wrong, or |
18 |
> misplaced, or the person writing them is overcautious to the point of being |
19 |
> ridiculous. If this advice really was true, then a whole lot of stuff would |
20 |
> break all over the world: |
21 |
> |
22 |
> - every Windows box on the planet would need a complete reinstall whenever a |
23 |
> Windows Update happened (Yes, Microsoft does upgrade their compiler!) |
24 |
> - third party apps would not run, as you have no way of knowing if Oracle's |
25 |
> compiler is the same as yours (and you don't even have a guarantee that Oracle |
26 |
> uses gcc). My Oracle instance at work is working just fine and I know for a |
27 |
> fact the compilers used for it and SuSE are not even in the same version |
28 |
> series. |
29 |
> - Compiling any package locally could not work on a binary distro. But they |
30 |
> do. |
31 |
> |
32 |
> There are *some* special cases where the gcc devs break stuff at an ABI level |
33 |
> between versions (usually related to C++ not to C). These are well known and |
34 |
> heavily documented - the toolchain devs make sure of this. 3.3 to 3.4 was such |
35 |
> a case, there was another minor case early in the gcc-4 series. By no means do |
36 |
> this mean that the fix for those cases must now be applied every time. |
37 |
> |
38 |
|
39 |
I must confess that I don't know if there is an ABI breakage between |
40 |
4.1.2 and 4.3.2. So if there is none you may be fine without rebuilding |
41 |
world. |
42 |
|
43 |
-- |
44 |
Daniel Pielmeier |