archive mirror
 help / color / mirror / Atom feed
From: Patrick Steinhardt <>
To: Junio C Hamano <>
Cc: Ghanshyam Thakkar <>,
	Achu Luma <>,,,
	Christian Couder <>,
	Johannes Schindelin <>
Subject: Re: [Outreachy][PATCH 2/2] Port helper/test-date.c to unit-tests/t-date.c
Date: Tue, 28 May 2024 15:20:13 +0200	[thread overview]
Message-ID: <ZlXaDWy0lQA1FM7d@tanuki> (raw)
In-Reply-To: <xmqqttkquxes.fsf@gitster.g>

[-- Attachment #1: Type: text/plain, Size: 2539 bytes --]

On Thu, Mar 28, 2024 at 09:35:39AM -0700, Junio C Hamano wrote:
> Ghanshyam Thakkar <> writes:
> > I believe the issue might not be related to the setenv function, but rather
> > with tzset(). As you can see here[1], when we set TZ before we call the
> > unit-testing binaries, the tests which were failing (EST5 ones that I
> > separated with t-datetest) pass on 'win test (0)', and the ones which
> > were passing (UTC ones, t-date) fail. (Although some tests on linux are also
> > failing, but that can be explained by the fact that t-date runs first
> > and sets the TZ to UTC, afterwhich t-datetest runs and fails, although
> > this is not conclusive). Therefore, I am almost certain that the issue
> > is with changing the timezone during runtime on windows and not with setting
> > TZ variable with setenv(). CC'ing Johannes to see if he has any insights
> > on this.
> Interesting.  Sometime before I started working on Git, I learned
> that no program did tzset() after it started running to switch
> multiple timezones and worked correctly on many different variants
> of UNIXes (there were many of them back then), and because I never
> got interested in writing a world-clock program, I didn't know, and
> kind of surprised to learn that it works on some platforms (like
> Linux and macOS) to switch zones with tzset() these days ;-).
> So, if Windows runtime is unhappy with the program calling tzset()
> more than once, I wouldn't be too surprised.
> Thanks.

As I was debugging other Windows-specific issues in a VM already, Chris
asked me to also have a look at this issue. And indeed, most of the
tests fail deterministically. I also found a fix:

    diff --git a/t/unit-tests/t-date.c b/t/unit-tests/t-date.c
    index dd5dbbb2e0..2d7b1f085a 100644
    --- a/t/unit-tests/t-date.c
    +++ b/t/unit-tests/t-date.c
    @@ -31,7 +31,7 @@ static int check_prereqs(unsigned int prereqs) {
     static void set_TZ_env(const char *zone) {
    -	setenv("TZ", zone, 1);
    +	_putenv_s("TZ", zone);

I have no idea why that works though, and the fix is of course not
portable. But with this change, the timezones do get picked up by
`tzset()` and related date functions as expected.

I'm quite dumb when it comes to the Windows API, so I don't have much of
a clue why this works. The documentation also didn't point out anything
obvious. Dscho, do you happen to have an explanation for this?


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

  reply	other threads:[~2024-05-28 13:20 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-02-05 16:25 [Outreachy][PATCH 1/2] date: refactor 64 bit prereq code into reusable functions Achu Luma
2024-02-05 16:25 ` [Outreachy][PATCH 2/2] Port helper/test-date.c to unit-tests/t-date.c Achu Luma
2024-03-28 12:35   ` Ghanshyam Thakkar
2024-03-28 16:35     ` Junio C Hamano
2024-05-28 13:20       ` Patrick Steinhardt [this message]
2024-05-28 16:41         ` Junio C Hamano
2024-05-29  5:49           ` Patrick Steinhardt
2024-06-27  6:49             ` Johannes Schindelin
2024-02-05 17:34 ` [Outreachy][PATCH 1/2] date: refactor 64 bit prereq code into reusable functions rsbecker
2024-02-06  8:39   ` Christian Couder

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=ZlXaDWy0lQA1FM7d@tanuki \ \ \ \ \ \ \ \ \

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).