Re: bug#54764: encode-time: make DST and TIMEZONE fields of the list argument optional ones

[Paul Eggert <eggert@cs.ucla.edu>]

[-- Attachment #1: Type: text/plain, Size: 1934 bytes --]

On 4/20/22 09:56, Max Nikulin wrote:

> A typo: double "a".

Thanks, fixed in the attached which I installed in master.


> If I get your idea correctly then "January, 31" + "1 month" should be 
> more impressive as impossible date.

Thanks, good idea; also in the attached patch.


> I can not figure out which elisp function can help to 
> determine wall time for Aug 1 start of day in Cairo:
> 
> Africa/Cairo  Thu Jul 31 21:59:59 2014 UT = Thu Jul 31 23:59:59 2014 EET 
> isdst=0 gmtoff=7200
> Africa/Cairo  Thu Jul 31 22:00:00 2014 UT = Fri Aug  1 01:00:00 2014 
> EEST isdst=1 gmtoff=10800
> 
> input: 2014-08-01 Africa/Cairo
> (timezone may be implicit as the system one)
> expected output: 01:00:00

Given mktime's limitations there's no trivial way to do this for 
arbitrary timestamps, since 00:00 doesn't exist in Cairo that day. 
Worse, in some locations near the International Date Line entire days do 
not exist, because at 00:00 they advanced the clocks forward 24 hours in 
order to move the date line.

It sounds like you're asking for a function that, given a date, yields 
the first broken-down timestamp on or after 00:00 of that date. For 
something like that, I'd use encode-time on 00:00 of that date to get a 
timestamp T, and then use time-add and decode-time to decode T-86400 
seconds, T, and T+86400 seconds, and if the decoded times all look fine 
then return (decode-time T). If not (i.e., their UTC offsets differ, or 
T's decoded time is not 00:00 on the correct date) I'd use binary search 
to find discontinuities between T-86400 and T+86400 and look next to 
those discontinuities to find timestamps closer to what you want.

Of course this is not ideal - but it's similar to what many mktime 
implementations do internally, and it's also similar to what Emacs's 
cal-dst already does (maybe you can look there for ideas), so you'd be 
in good company.

[-- Attachment #2: 0001-More-encode-time-pitfall-doc-fixes.patch --]
[-- Type: text/x-patch, Size: 2651 bytes --]

From f98c3f4426fecf794f47f27aebe1f3b854fb1bfd Mon Sep 17 00:00:00 2001
From: Paul Eggert <eggert@cs.ucla.edu>
Date: Wed, 20 Apr 2022 12:03:19 -0700
Subject: [PATCH] More encode-time pitfall doc fixes

* doc/lispref/os.texi (Time Conversion): Improve discussion of
encode-time pitfalls based on comments by Max Nikulin (Bug#54764#63).
---
 doc/lispref/os.texi | 22 ++++++++++++++--------
 1 file changed, 14 insertions(+), 8 deletions(-)

diff --git a/doc/lispref/os.texi b/doc/lispref/os.texi
index cabae08970..4138dab09f 100644
--- a/doc/lispref/os.texi
+++ b/doc/lispref/os.texi
@@ -1670,7 +1670,7 @@ Time Conversion
 convention, @var{dst} is @minus{}1 and @var{zone} defaults to the
 current time zone rule (@pxref{Time Zone Rules}).
 When modernizing an obsolescent caller, ensure that the more-modern
-list equivalent contains 9 elements with a a @code{dst} element that
+list equivalent contains 9 elements with a @code{dst} element that
 is @minus{}1, not @code{nil}.
 
 Year numbers less than 100 are not treated specially.  If you want them
@@ -1695,22 +1695,28 @@ Time Conversion
 For example:
 
 @lisp
-;; Try to compute the time four years from now.
+;; Try to compute the time one month from now.
 ;; Watch out; this might not work as expected.
 (let ((time (decode-time)))
-  (setf (decoded-time-year time)
-        (+ (decoded-time-year time) 4))
+  (setf (decoded-time-month time)
+        (+ (decoded-time-month time) 1))
   time)
 @end lisp
 
 @noindent
 Unfortunately, this code might not work as expected if the resulting
-time is invalid due to daylight saving transitions, time zone changes,
+time is invalid due to month length differences,
+daylight saving transitions, time zone changes,
 or missing leap days or leap seconds.  For example, if executed on
-February 29, 2096 this code yields a nonexistent date because 2100 is
-not a leap year.  To avoid some (though not all) of the problem, you
+January 30 this code yields a nonexistent date February 30,
+which @code{encode-time} would adjust to early March.
+Similarly, adding four years to February 29, 2096 would yield the
+nonexistent date February 29, 2100; and adding one hour to 01:30 on
+March 13, 2022 in New York would yield a timestamp 02:30 that does not
+exist because clocks sprang forward from 02:00 to 03:00 that day.
+To avoid some (though not all) of the problem, you
 can base calculations on the middle of the affected unit, e.g., start
-at July 1 when adding years.  Alternatively, you can use the
+at the 15th of the month when adding months.  Alternatively, you can use the
 @file{calendar} and @file{time-date} libraries.
 @end defun
 
-- 
2.32.0