bug#59501: [PATCH] rcirc: Issue with printing messages in the wrong place

[Philip Kaludercic <philipk@posteo.net>]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: Philip Kaludercic <philipk@posteo.net>
>> Cc: 59501@debbugs.gnu.org, "'Eli Zaretskii'" <eliz@gnu.org>
>> Date: Mon, 01 May 2023 11:59:57 +0000
>> 
>> Thuna <thuna.cing@gmail.com> writes:
>> 
>> > I have occasionally had replies show before my messages, so I decided
>> > to look into it.   It looks like rcirc looks at the time of the
>> > message, compares it to the `rcirc-time' text property of other
>> > messages, and places it so that the property is properly ordered.  >
>> > ============================= Disregard this, the patch does not > fix
>> > it.  I decided to keep it in because the problem is still > possible,
>> > but not it's not what's responsible for this specific > issue. The
>> > issue with this is, because the recieved messages' precision can be
>> > less than the sent messages' precision (`current-time' returns the
>> > time down to milliseconds but servers may choose to only keep it down
>> > to seconds), this causes some issues when a message is recieved after
>> > a message was sent but still within the same second. >
>> > =============================  An example of this, with the value of
>> > the text property `rcirc-time' at the left of the message itself, is:
>> >> (25469 29640 874652 270000)   > 02:13 <thuna`> ,ping
>> >> (25469 29641)                 > 02:13 <fsbot> ␕
>> >> (25469 29642 502966 649000)   > 02:13 <thuna`> ,ping
>> >> (25469 29643)                 > 02:13 <fsbot> ␕
>> >> (25469 29643 755653 540000)   > 02:13 <thuna`> ,ping
>> >> (25469 29644)                 > 02:13 <fsbot> ␕
>> >> (25469 29645)                 > 02:13 <fsbot> ␕
>> >> (25469 29645 103860 988000)   > 02:13 <thuna`> ,ping
>> >
>> > As you can see, the response to the last ping arrived after the ping
>> > itself, but because of the precision difference, rcirc sorted the
>> > messages wrong.
>> >
>> > The issue was with `parse-time-string'.  It (and by proxy
>> > `parse-iso8601-time-string'), uses `iso8601-parse' with nil FORM,
>> > resulting in the subsecond precision being thrown out.  To fix this, I
>> > added FORM arguments to those two functions as well.
>> 
>> Sorry for missing this message, this patch looks like the right fix.  I
>> have recently been noticing issues related to this myself.
>> 
>> Would it be OK to apply this to emacs-29?
>
> Yes, but please fix the doc string to not use non-ASCII quotes.
>
> Also, don't forget the Copyright-paperwork-exempt thingy when you
> install.

I've pushed the change, closing this report.