Re: [PATCH] daemon: restore getpeername(0,...) use

["Joachim Schmitz" <jojo@schmitz-digital.de>]

Jeff King wrote:
> On Sat, Sep 08, 2012 at 09:20:48PM +0200, Jan Engelhardt wrote:
>
>> On Saturday 2012-09-08 20:59, Junio C Hamano wrote:
>>>> diff --git a/daemon.c b/daemon.c
>>>> index 4602b46..eaf08c2 100644
>>>> --- a/daemon.c
>>>> +++ b/daemon.c
>>>> @@ -1,3 +1,4 @@
>>>> +#include <stdbool.h>
>>>>  #include "cache.h"
>>>>  #include "pkt-line.h"
>>>>  #include "exec_cmd.h"
>>>
>>> Platform agnostic parts of the code that use "git-compat-util.h"
>>> (users of "cache.h" are indirectly users of it) are not allowed to
>>> do platform specific include like this at their beginning.
>>>
>>> This is the first use of stdbool.h; what do you need it for?
>>
>> For the use in setenv(,,true). It was not entirely obvious in which
>> .h to add it; the most reasonable place was daemon.c itself, since
>> the other .c files do not seem to need it.
>
> It would go in git-compat-util.h. However, it really is not needed;
> you can simply pass "1" to setenv, as every other callsite in git
> does.
>
> More importantly, though, is it actually portable? I thought it was
> added in C99, and we try to stick to C89 to support older compilers
> and systems. My copy of C99 is vague (it says only that the "bool"
> macro was added via stdbool.h in C99, but nothing about the "true"
> and "false" macros), and I don't have a copy of C89 handy.  Wikipedia
> does claim the header wasn't standardized at all until C99:
>
>  https://en.wikipedia.org/wiki/C_standard_library

Indeed stdbool is not part of C89, but inline isn't either and used 
extensively in git (could possible be defined away), as are non-const array 
intializers, e.g.:


                const char *args[] = { editor, path, NULL };
                                               ^
".../git/editor.c", line 39: error(122): expression must have a constant 
value

So git source is not plain C89 code (anymore?)

Bye, Jojo