GCC >= 5? (was: [patch] Removal of strfwd.h)

[Jean-Marc Lasgouttes]

Le 04/10/2024 à 17:20, Pavel Sanda a écrit :
> On Fri, Oct 04, 2024 at 04:46:01PM +0200, Jean-Marc Lasgouttes wrote:
>> I did not read completely your comment in docstring.h. I think we can get
>> rid completely of the names trvstring and trivdocstring soon after this
>> landed.
> 
> If we are sure that the thread safety issue is gone for all reasonable
> STL implemtations then yes (so far I saw only explicit ack for gcc5,
> but did not googled around).

This threadsafe string implementation is a requirement for C++11 AFAIK.

Have a look at the last row of the last table here:
https://en.cppreference.com/w/cpp/11

>> In this view, I do not think that we need a test to check that the
>> c++ implementation of string works, do we?
> 
> I meant is as a check for our docstring wrapper. But I don't have
> any skin in this game and am completely fine with removing it.

docstring is basically basic_string<our_wide_char>

If there are issues, this may be with our wide char choice (hopefully 
wchar_t). The code there is very scary, but I do not know whether it is 
still needed.

JMarc