# [PATCH] Fix date() bug - Thread: https://externals.io/message/40726 - Original archive: https://news-web.php.net/php.internals/40726 - Started by Ilia Cheishvili on 2008-09-27T07:04:34+00:00 - Messages: 9 - Reply by email to internals@lists.php.net - More for agents: search with https://externals.io/search.md?q={query} · JSON API at https://externals.io/api · documentation at https://externals.io/llms.txt --- ## #40726 — Ilia Cheishvili — 2008-09-27T07:04:34+00:00 URL: https://externals.io/message/40726#40726 Hi all, This patch addresses the issue with the `date()` function. When passing in a 'u', the `date()` function simply outputs six zeros. To fix this, I added a `gettimeofday()` call that figures out what to display for microseconds. I am including the headers and using the function with pre-processor safeguards as well. Take a look :) Thanks! Ilia --- ## #40727 — Alexey Zakhlestin — 2008-09-27T07:40:52+00:00 URL: https://externals.io/message/40726#40727 · In reply to: #40726 On Sat, Sep 27, 2008 at 11:04 AM, Ilia Cheishvili wrote: > Hi all, > This patch addresses the issue with the `date()` function. When passing in a > 'u', the `date()` function simply outputs six zeros. To fix this, I added a > `gettimeofday()` call that figures out what to display for microseconds. I am > including the headers and using the function with pre-processor safeguards > as well. > Take a look :) Wouldn't it be better, to make `gettimeofday()` call only in case of 'u'? -- Alexey Zakhlestin http://blog.milkfarmsoft.com/ --- ## #40730 — Ilia Cheishvili — 2008-09-27T08:04:03+00:00 URL: https://externals.io/message/40726#40730 · In reply to: #40727 It definitely would be, and that's actually the way I would have preferred to do it. I didn't want to impact too much code, if that makes sense in this case, but I'm glad that someone agrees :) I have attached a patch to do exactly this. Ilia On Sat, Sep 27, 2008 at 1:40 AM, Alexey Zakhlestin wrote: > On Sat, Sep 27, 2008 at 11:04 AM, Ilia Cheishvili > wrote: > > […] > a > > […] > a > > […] > am > > […] > safeguards > > […] > > Wouldn't it be better, to make `gettimeofday()` call only in case of 'u'? > > > -- > Alexey Zakhlestin > http://blog.milkfarmsoft.com/ --- ## #40731 — Alexey Zakhlestin — 2008-09-27T08:20:42+00:00 URL: https://externals.io/message/40726#40731 · In reply to: #40730 On Sat, Sep 27, 2008 at 12:04 PM, Ilia Cheishvili wrote: > It definitely would be, and that's actually the way I would have preferred > to do it. I didn't want to impact too much code, if that makes sense in > this case, but I'm glad that someone agrees :) > I have attached a patch to do exactly this. I was thinking other in another direction.. case 'u': #ifdef HAVE_GETTIMEOFDAY gettimeofday(&tp, &tz); length = slprintf(buffer, 32, "%06d", (int) tp.tv_usec); #else length = slprintf(buffer, 32, "%06d", (int) floor(t->f * 1000000)); #endif break; > > On Sat, Sep 27, 2008 at 1:40 AM, Alexey Zakhlestin > wrote: > > […] > > -- Alexey Zakhlestin http://blog.milkfarmsoft.com/ --- ## #40732 — Ilia Cheishvili — 2008-09-27T08:26:58+00:00 URL: https://externals.io/message/40726#40732 · In reply to: #40731 Ah, I see. I like that even better :) And I attached yet another patch that incorporates your idea. Ilia On Sat, Sep 27, 2008 at 2:20 AM, Alexey Zakhlestin wrote: > On Sat, Sep 27, 2008 at 12:04 PM, Ilia Cheishvili > wrote: > > […] > preferred > > […] > > I was thinking other in another direction.. > > case 'u': > #ifdef HAVE_GETTIMEOFDAY > gettimeofday(&tp, &tz); > length = slprintf(buffer, 32, "%06d", (int) tp.tv_usec); > #else > length = slprintf(buffer, 32, "%06d", (int) floor(t->f * 1000000)); > #endif > break; > > > > […] > added > > […] > > > > -- > Alexey Zakhlestin > http://blog.milkfarmsoft.com/ --- ## #40734 — Alexey Zakhlestin — 2008-09-27T09:21:18+00:00 URL: https://externals.io/message/40726#40734 · In reply to: #40732 On Sat, Sep 27, 2008 at 12:26 PM, Ilia Cheishvili wrote: > Ah, I see. I like that even better :) > And I attached yet another patch that incorporates your idea. thanks. I did some more code-digging, and it looks, like proper point for fixing is not here, anyway. It should be in timelib_unixtime2gmt() and timelib_unixtime2local() funcitons of ext/date/lib/unixtime2tm.c "->f" member of structure is not set there, which results in zero's in output. `date()` function is an abstraction and should not deal with such details itself. > > On Sat, Sep 27, 2008 at 2:20 AM, Alexey Zakhlestin > wrote: > > […] > > -- Alexey Zakhlestin http://blog.milkfarmsoft.com/ --- ## #40735 — Hannes Magnusson — 2008-09-27T09:28:35+00:00 URL: https://externals.io/message/40726#40735 · In reply to: #40726 On Sat, Sep 27, 2008 at 09:04, Ilia Cheishvili wrote: > Hi all, > This patch addresses the issue with the `date()` function. When passing in a > 'u', the `date()` function simply outputs six zeros. To fix this, I added a "Note: Since this function only accepts integer timestamps the u format character is only useful when using the `date_format()` function with user based timestamps created with `date_create()`. " ? -Hannes --- ## #40736 — Alexey Zakhlestin — 2008-09-27T09:43:55+00:00 URL: https://externals.io/message/40726#40736 · In reply to: #40735 On Sat, Sep 27, 2008 at 1:28 PM, Hannes Magnusson wrote: > On Sat, Sep 27, 2008 at 09:04, Ilia Cheishvili > wrote: > > […] > > "Note: Since this function only accepts integer timestamps the u > format character is only useful when using the `date_format()` function > with user based timestamps created with `date_create()`. " ? This is oversimplification, as for "current moment", `date()` can get microseconds. For user-provided timestamps it, obviously, can not — that is true. -- Alexey Zakhlestin http://blog.milkfarmsoft.com/ --- ## #40739 — Ilia Cheishvili — 2008-09-27T21:05:08+00:00 URL: https://externals.io/message/40726#40739 · In reply to: #40734 I did some digging around in the code and came up with the solution that you suggested. I put the fix in unixtime2tm.c, letting php_date.c remain the nice abstraction layer between the system-dependent code of getting the time and actually formatting it for display. The patch is attached. Thanks for the back and forth thus far :) Ilia On Sat, Sep 27, 2008 at 3:21 AM, Alexey Zakhlestin wrote: > On Sat, Sep 27, 2008 at 12:26 PM, Ilia Cheishvili > wrote: > > […] > > thanks. > I did some more code-digging, and it looks, like proper point for > fixing is not here, anyway. > > It should be in timelib_unixtime2gmt() and timelib_unixtime2local() > funcitons of ext/date/lib/unixtime2tm.c > "->f" member of structure is not set there, which results in zero's in > output. > > `date()` function is an abstraction and should not deal with such details > itself. > > > […] > in > > […] > indeyets@gmail.com> > > […] > 'u'? > > […] > > > > -- > Alexey Zakhlestin > http://blog.milkfarmsoft.com/