# Optional nowait argument to sem_acquire - Thread: https://externals.io/message/77035 - Original archive: https://news-web.php.net/php.internals/77035 - Started by Matteo Beccati on 2014-09-02T12:50:29+00:00 - Messages: 8 - Reply by email to internals@lists.php.net - More for agents: search with https://externals.io/search.md?q={query} · RFCs at https://externals.io/rfc.md · JSON API at https://externals.io/api · documentation at https://externals.io/llms.txt --- ## #77035 — Matteo Beccati — 2014-09-02T12:50:29+00:00 URL: https://externals.io/message/77035#77035 Hi everyone, the following patch: https://github.com/php/php-src/pull/741 has been lying around somewhere on my box since years and I recently stumbled upon it. However 5.6 was already RC and Ferenc suggested to eventually wait for 5.6.1. Do you think it's ok to add such a minor feature to 5.6.1? Cheers -- Matteo Beccati Development & Consulting - http://www.beccati.com/ --- ## #77037 — guilhermeblanco@gmail.com — 2014-09-02T13:38:37+00:00 URL: https://externals.io/message/77035#77037 · In reply to: #77035 Hi, Since I don't know where you guys like to discuss implementations, I'm pasting same thing I did in php-src PR: I'd propose instead of accepting "nowait" only, I'd use a better support and accept "until", dealing with possible cases: - -1: Blocks and wait until lock is acquired - 0: Exactly what your patch does, attempt to grab and if it fails, just return false (I'd remove the E_WARNING) - > 0: Acceptable time in milliseconds it would wait until consider the acquire failed Cheers, > Hi everyone, > > the following patch: > > https://github.com/php/php-src/pull/741 > > has been lying around somewhere on my box since years and I recently > stumbled upon it. > > However 5.6 was already RC and Ferenc suggested to eventually wait for > 5.6.1. > > Do you think it's ok to add such a minor feature to 5.6.1? > > > Cheers > -- > Matteo Beccati > > Development & Consulting - http://www.beccati.com/ > > -- > > > > -- Guilherme Blanco MSN: guilhermeblanco@hotmail.com GTalk: guilhermeblanco Toronto - ON/Canada --- ## #77038 — Matteo Beccati — 2014-09-02T14:01:04+00:00 URL: https://externals.io/message/77035#77038 · In reply to: #77037 Hi, > I'd propose instead of accepting "nowait" only, I'd use a better support > and accept "until", dealing with possible cases: > > * -1: Blocks and wait until lock is acquired > * 0: Exactly what your patch does, attempt to grab and if it fails, > just return false (I'd remove the E_WARNING) (It doesn't. The warning is raised when the semaphore can't be acquired for other reasons) > * > 0: Acceptable time in milliseconds it would wait until consider > the acquire failed TBH, I don't know. semtimedop is not universally available (no BSD, as far as I can tell) and I would prefer to avoid writing an alternative implementation in case it's not available. On the other hand triggering an error in such case wouldn't be nice. Cheers -- Matteo Beccati Development & Consulting - http://www.beccati.com/ --- ## #77039 — guilhermeblanco@gmail.com — 2014-09-02T14:39:02+00:00 URL: https://externals.io/message/77035#77039 · In reply to: #77038 Hi Matteo, BSD does accept it as sem_timedwait, which I could find a reference for FreeBSD: http://www.freebsd.org/cgi/man.cgi?query=sema&sektion=9 Cheers, > Hi, > > > > […] > > (It doesn't. The warning is raised when the semaphore can't be acquired > for other reasons) > > > […] > > TBH, I don't know. semtimedop is not universally available (no BSD, as > far as I can tell) and I would prefer to avoid writing an alternative > implementation in case it's not available. On the other hand triggering > an error in such case wouldn't be nice. > > > Cheers > -- > Matteo Beccati > > Development & Consulting - http://www.beccati.com/ > -- Guilherme Blanco MSN: guilhermeblanco@hotmail.com GTalk: guilhermeblanco Toronto - ON/Canada --- ## #77040 — Matteo Beccati — 2014-09-02T14:46:56+00:00 URL: https://externals.io/message/77035#77040 · In reply to: #77039 > Hi Matteo, > > BSD does accept it as sem_timedwait, which I could find a reference for > FreeBSD: > > http://www.freebsd.org/cgi/man.cgi?query=sema&sektion=9 that's part of the FreeBSD Kernel Developer's Manual, not SysV semaphores ;) Cheers -- Matteo Beccati Development & Consulting - http://www.beccati.com/ --- ## #77041 — guilhermeblanco@gmail.com — 2014-09-02T15:03:43+00:00 URL: https://externals.io/message/77035#77041 · In reply to: #77040 It's also supported by System V using exactly same API. Some references: http://fossies.org/dox/glibc-2.19/sem__timedwait_8c_source.html http://manned.org/sem_timedwait/83baa20c http://linux.die.net/man/3/sem_timedwait http://pubs.opengroup.org/onlinepubs/009695399/functions/sem_timedwait.html > > > […] > > that's part of the FreeBSD Kernel Developer's Manual, not SysV semaphores > ;) > > > Cheers > -- > Matteo Beccati > > Development & Consulting - http://www.beccati.com/ > -- Guilherme Blanco MSN: guilhermeblanco@hotmail.com GTalk: guilhermeblanco Toronto - ON/Canada --- ## #77073 — guilhermeblanco@gmail.com — 2014-09-05T15:38:16+00:00 URL: https://externals.io/message/77035#77073 · In reply to: #77041 After discussing with Matteo, I realized I was referring to POSIX sempahores and not to System V. Patch looks ok to me. Is it possible to have this merged? Thanks, On Tue, Sep 2, 2014 at 11:03 AM, guilhermeblanco@gmail.com < guilhermeblanco@gmail.com> wrote: > It's also supported by System V using exactly same API. > Some references: > > http://fossies.org/dox/glibc-2.19/sem__timedwait_8c_source.html > http://manned.org/sem_timedwait/83baa20c > http://linux.die.net/man/3/sem_timedwait > http://pubs.opengroup.org/onlinepubs/009695399/functions/sem_timedwait.html > > > > > > […] > > > > -- > Guilherme Blanco > MSN: guilhermeblanco@hotmail.com > GTalk: guilhermeblanco > Toronto - ON/Canada > -- Guilherme Blanco MSN: guilhermeblanco@hotmail.com GTalk: guilhermeblanco Toronto - ON/Canada --- ## #77132 — Matteo Beccati — 2014-09-10T07:30:35+00:00 URL: https://externals.io/message/77035#77132 · In reply to: #77073 > After discussing with Matteo, I realized I was referring to POSIX > sempahores and not to System V. > > Patch looks ok to me. > > Is it possible to have this merged? I've seen no objections raised, so I committed it to PHP-5.6 and master. Cheers -- Matteo Beccati Development & Consulting - http://www.beccati.com/