apr-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Yann Ylavic <ylavic....@gmail.com>
Subject Re: svn commit: r1797416 - in /apr/apr/branches/1.7.x: ./ locks/beos/proc_mutex.c locks/os2/proc_mutex.c locks/win32/proc_mutex.c
Date Sat, 03 Jun 2017 18:13:19 GMT
On Sat, Jun 3, 2017 at 8:07 PM, Yann Ylavic <ylavic.dev@gmail.com> wrote:
> On Sat, Jun 3, 2017 at 7:15 PM, William A Rowe Jr <wrowe@rowe-clan.net> wrote:
>> On Jun 3, 2017 10:53, "Yann Ylavic" <ylavic.dev@gmail.com> wrote:
>>
>>
>> Right, but why apr_os_mutex_set[_ex](DEFAULT_TIMED) should fail on
>> Windows, OS2 and BEOS?
>> APR_LOCK_DEFAULT_TIMED = APR_LOCK_DEFAULT allows this, and does not
>> break with _get[_ex]() either I think.
>>
>>
>> How do you mean, it is allowed right now with the current implementations?
>
> APR_DECLARE(apr_status_t) apr_os_proc_mutex_put_ex(apr_proc_mutex_t **pmutex,
>                                                 apr_os_proc_mutex_t *ospmutex,
>                                                 apr_lockmech_e mech,
>                                                 int register_cleanup,
>                                                 apr_pool_t *pool);
>
> It takes a mech as argument, the previous implementation allowed to
> specify DEFAULT_TIMED not only for unixes, but also for Windows, OS2
> and BEOS.
> But now it would fail, which does not really help user code portability IMHO.

Oh, actually you didn't change _put_ex(), I misread your commit.
Please ignore this ;)

Mime
View raw message