Martin Storsjö <[email protected]> writes:

> On Sat, 13 Oct 2012, Hendrik Leppkes wrote:
>
>> On Sat, Oct 13, 2012 at 8:14 PM, Måns Rullgård <[email protected]> wrote:
>>> Hendrik Leppkes <[email protected]> writes:
>>>
>>>> On Sat, Oct 13, 2012 at 7:46 PM, Derek Buitenhuis
>>>> <[email protected]> wrote:
>>>>> On 13/10/2012 1:29 PM, Måns Rullgård wrote:
>>>>>> I don't really like calling the OS msvc.  That is the compiler.  How
>>>>>> about "windows"?
>>>>>
>>>>> By that logic, shouldn't we also call mingw32 "windows"?
>>>>>
>>>>
>>>> Yeah, technically its the same OS, and the configuration how to create
>>>> shared libraries doesn't belong in a OS section, but a toolchain
>>>> section - but thats being done for like every OS, because in most
>>>> cases there is an assumption that there is only one wide-spread
>>>> compiler/linker for that target OS.
>>>
>>> The way (shared) libraries are built is usually dictated much more by
>>> the OS than by the specific compiler.  It is, after all, the OS that in
>>> the end will be loading them.  Windows is the odd one out here with at
>>> least three totally different schemes in common use.
>>>
>>> If you look at the other ones, the settings are very much per OS and
>>> hardly per compiler at all.  Symbian is a prime example, building with
>>> gcc yet needing a raft of special flags.
>>>
>>>> In any case, i would suggest going with something as simple as
>>>> "windows" or "win32" to stick with microsofts short-form :p
>>>
>>> I'm fine with win32 as well.  Is the 64-bit Windows also called win32?
>>>
>>
>> Win64 is used sometimes when you want to stress the difference, but
>> its the same OS in 64-bit, does it make sense to separate here? Maybe
>> stick to "windows" and keep the 32/64 out of it to avoid confusion?
>
> Some of the target_os cases have * to match different settings, I can
> set it to win*.

That sounds a bit greedy.  Matching win32|win64 should be safe.

-- 
Måns Rullgård
[email protected]
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to