Gravity: Sorting by date Google Group posts

[Steve]

I sort my messages in Gravity by date rather than by the
subject. It makes it easier for me to see the newest
messages. Google Groups posts aren't in order because of
the timezone. Is there a way to make Gravity sort GG
posts by 'Injection-Date' instead of 'Date'? Thanks.

[Marcel Logen]

Steve in news.software.readers:

I don't know Gravity, but perhaps there is a possibility
to sort by (server) article number?

Marcel
--
╭───╮ ╭─╮ ╭────────╮ ╭─╮ ╭───╮ ╭──────╮
───╯ │ │ ╰──╮ ╰────╮ ╰─╯ │ │ ╰───╮ ╭─╮ ╰────╮ │ ╭─
╰─╮ ╰─╮ ╰──╮ ╭──╯ ╭────╯ ╰────╮ ╰─╮ │ │ ╭─╮ ╭──╯ │ │
╰─────────╯ ╰─╯ ╰────────────╯ ╰─╯ ╰──╯ ╰─╯ ╰──╯

[Frank Slootweg]

I've only looked at a few examples, but AFAICT, 'Date:' and
'Injection-Date:' (of Google Groups posts) contain the same time.

'Date:' contains the user's (not the server's/Google's) local time,
with a timezone code. 'Injection-Date:' also contains the user's local
time, but converted to UTC (with '+0000' at the end).

If you think 'Date: and 'Injection-Date:' are different, then give the
message-id or/and relevant headers of an example.

[Michael Bäuerle]

Frank Slootweg wrote:
>
> [...]

> I've only looked at a few examples, but AFAICT, 'Date:' and
> 'Injection-Date:' (of Google Groups posts) contain the same time.
>
> 'Date:' contains the user's (not the server's/Google's) local time,
> with a timezone code. 'Injection-Date:' also contains the user's local
> time, but converted to UTC (with '+0000' at the end).

This is common, but not mandatory.

> If you think 'Date: and 'Injection-Date:' are different, then give the
> message-id or/and relevant headers of an example.

One of my articles: <news:rs0pvt$k25$1...@wieslauf.sub.de>
|
| [...]
| Date: Thu, 24 Dec 2020 02:13:33 +0100
| [...]
| Injection-Date: Thu, 24 Dec 2020 01:13:34 -0000 (UTC)

But a larger difference and non-UTC timezone for Injection-Date would
be allowed too.

RFC 5536 specify date-time for the content of both header fields:
<https://tools.ietf.org/html/rfc5536#section-3.1.1>
<https://tools.ietf.org/html/rfc5536#section-3.2.7>

date-time is defined by RFC 5322:
<https://tools.ietf.org/html/rfc5322#section-3.3>

[Steve]

On Fri, 1 Jan 2021 18:39:51 +0100, Marcel Logen wrote:
>
> Steve in news.software.readers:
>
> >I sort my messages in Gravity by date rather than by the
> >subject. It makes it easier for me to see the newest
> >messages. Google Groups posts aren't in order because of
> >the timezone. Is there a way to make Gravity sort GG
> >posts by 'Injection-Date' instead of 'Date'? Thanks.
>
> I don't know Gravity, but perhaps there is a possibility
> to sort by (server) article number?
>
> Marcel

Thanks, I had a look, but I can't find a setting for that
in Gravity.

[Steve]

Example header.

X-Received: by 2002:a0c:fdec:: with SMTP id
m12mr65695485qvu.11.1609500095042; Fri, 01 Jan
2021 03:21:35 -0800 (PST)
X-Received: by 2002:ac8:5286:: with SMTP id
s6mr60318561qtn.22.1609500094851; Fri, 01 Jan 2021
03:21:34 -0800 (PST)
Path: not-for-mail
Newsgroups: uk.telecom.mobile
Date: Fri, 1 Jan 2021 03:21:34 -0800 (PST) <----
User-Agent: G2/1.0
MIME-Version: 1.0
Message-ID: <3be7b521-45b4-4b71-89fc-
d39d68...@googlegroups.com>
Subject: Re: EU roaming charges will be coming back
From: "notya...@gmail.com" <notya...@gmail.com>
Injection-Date: Fri, 01 Jan 2021 11:21:35 +0000 <----

And this is how I see it in Gravity

Thread pane showing article out of date order
<https://ibb.co/Ypyxk63>

Article headers
<https://ibb.co/5TpxWv6>

40tude Dialog shows the messages in correct order, but I
prefer Gravity. I've gone through the settings in Gravity
many times and can't find any way to change this.

[Frank Slootweg]

Michael Bäuerle <michael....@gmx.net> wrote:
> Frank Slootweg wrote:
> >
> > [...]
> > I've only looked at a few examples, but AFAICT, 'Date:' and
> > 'Injection-Date:' (of Google Groups posts) contain the same time.
> >
> > 'Date:' contains the user's (not the server's/Google's) local time,
> > with a timezone code. 'Injection-Date:' also contains the user's local
> > time, but converted to UTC (with '+0000' at the end).
>
> This is common, but not mandatory.

I know, I just described what I saw for the few Google Groups posts
which I looked at.

> > If you think 'Date: and 'Injection-Date:' are different, then give the
> > message-id or/and relevant headers of an example.
>
> One of my articles: <news:rs0pvt$k25$1...@wieslauf.sub.de>
> |
> | [...]
> | Date: Thu, 24 Dec 2020 02:13:33 +0100
> | [...]
> | Injection-Date: Thu, 24 Dec 2020 01:13:34 -0000 (UTC)

Yes, these are the same time (only 1 second off). (Your time zone is 1
hour late off (after) UTC ('+0100'), so to get UTC subtract 1 hour from
02:13:33.)

> But a larger difference and non-UTC timezone for Injection-Date would
> be allowed too.

Yes, I know, but we're discussing Google Groups posts. For such posts
that shouldn't happen, because they're posted on-line / in real time,
not offline (where indeed Injection-Date could be (much) later than
Date).

[Frank Slootweg]

Seems OK. 03:21:34 local PST time, which is 8 hours early (-0800) of
UTC. So 03:21:34 + 8 hours to get UTC, which is listed as 11:21:35, i.e.
1 second off (which is of course OK).

> And this is how I see it in Gravity
>
> Thread pane showing article out of date order
> <https://ibb.co/Ypyxk63>

This thread pane seems to imply you sort by the local time of the
poster instead of your local time, because the 'problem' article is
listed as '01/1/2021 03:21' i.e. the poster's local time (unless you
live in the PST time zone (I can't tell from your headers, because
they're in UTC)).

Sorting by poster-local-time is unusual, because that *will* result in
out of date/time order, unless everybody is in the same timezone.

Anyway, the 'problem' article is not listed as out of (reverse, newest
first) date/time order ('31/12/2020 22;25' -> '01/1/2021 03:21' ->
'01/1/2021 08:04' (bottom to top) is the correct order).

Bottom line: I don't understand the problem you're having.

When responding please give your local time zone (and offset from UTC)
and on which time/timezone you sort in Gravity.

[Steve]

The problem as I see it is Google puts their timezone
(PST) into the 'Date' header and adds an 'Injection-
Date' header with the posters timezone which Gravity
ignores.
I was hoping a Gravity user may be able to help me,
because this problem I'm having with Gravity displaying
Google Group posts not in date order only happens in
Gravity. Thunderbird, 40tude and Xnews don't put the
Google posts out of order when sorting the articles by
date.

> When responding please give your local time zone (and offset from UTC)

(UTC+00:00) Dublin, Edinburgh, Lisbon, London

> and on which time/timezone you sort in Gravity.

It's the default system time and timezone. There is no
way to change this in Gravity.

[Frank Slootweg]

If 'notyalckram' is a UK poster, that's indeed a silly thing to do for
Google.

> and adds an 'Injection-
> Date' header with the posters timezone which Gravity
> ignores.

The 'Injection-Date' header is indeed reported in UTC, but that does
not mean that the poster lives in the UTC timezone. My 'Date' header is
also in UTC ('GMT'), but I live in The Netherlands, which (now, no DST)
is UTC+01:00.

All in all, I think that as - as I described before - 'Date' and
'Injection-Date' contain the exact same time, it looks like Gravity can
not handle the syntactically correct 'Date' header. That's rather
strange, because using a timezone code like '(PST)' is long standing
common practice.

Maybe Michael can see if there's anything wrong with the 'Date' header
(which I don't see).

Other than that, you might want to check if there are any other -
non-GG - articles with a '(XYZ)' timezone in their 'Date' header and see
whether Gravity can correctly handle those.

> I was hoping a Gravity user may be able to help me,
> because this problem I'm having with Gravity displaying
> Google Group posts not in date order only happens in
> Gravity. Thunderbird, 40tude and Xnews don't put the
> Google posts out of order when sorting the articles by
> date.

Exactly. That's why I suspect this is a Gravity bug, albeit a very
strange/basic one.

> > When responding please give your local time zone (and offset from UTC)
>
> (UTC+00:00) Dublin, Edinburgh, Lisbon, London

Thanks.

[Adam H. Kerman]

Steve <lamma...@gmail.com> wrote:

>>>Date: Fri, 1 Jan 2021 03:21:34 -0800 (PST) <----

>>>Injection-Date: Fri, 01 Jan 2021 11:21:35 +0000 <----

>The problem as I see it is Google puts their timezone
>(PST) into the 'Date' header

No, it hasn't. (PST) is a comment to contain a common human-readable
abbreviation for Pacific Standard Time. A comment is ignored.

As the OP told you, the time zone is expressed as an offset to Greenwich.
By expressing the time zone as an offset, all time calculations are based
on the time at Greenwich, no matter if a geographically-irrelevant time
zone offset is used.

In any event, -0800/-0700 is used because Google Groups is a Web
interface to the client. The client isn't on the user's machine but on
the same host as the server. Wherever in the world the server farm truly
is, -0800/-0700 is local time at Google headquarters. I don't believe
there's a setting in Google Groups for the user's local time.

>and adds an 'Injection- Date' header with the posters timezone which Gravity
>ignores.

It did no such thing. +0000 is used when the date in the header does not
state local time. It expressed Date-Time as UTC with no offset.

Greenwich Mean Time as local time in the UK is express with an offset
of -0000.

>. . .

>> When responding please give your local time zone (and offset from UTC)

>(UTC+00:00) Dublin, Edinburgh, Lisbon, London

Wrong. Please read RFC 2822. If GMT is local time, then it's -0000.
Because it will screw up parsing on a Date header, the colon as time
separator is never used.

>>. . .

[Steve]

Gravity is no longer maintained and has its annoyances,
but I prefer to the other newsreaders I've tried. This
google date thing and not being able to search by
message-id are the two that bug me the most.

> > > When responding please give your local time zone (and offset from UTC)
> >
> > (UTC+00:00) Dublin, Edinburgh, Lisbon, London
>
> Thanks.

Thanks for taking the time to look at this, Frank.

[Steve]

On Sun, 3 Jan 2021 21:14:17 -0000 (UTC), Adam H. Kerman
wrote:
[...]
> Wrong.

Don't you have anything better to do than being a prize
prick, Adam?

[Adam H. Kerman]

Adam H. Kerman <a...@chinet.com> wrote:

>It did no such thing. +0000 is used when the date in the header does not
>state local time. It expressed Date-Time as UTC with no offset.

You know, I got it backwards and I'm not being helpful. I apologize.

If local time is GMT or any location in which UTC is local time, the
offset is expressed as +0000. If time is expressed as UTC without regard
to local time, then the offset is -0000. See RFC 2822.

[Adam H. Kerman]

Frank Slootweg <th...@ddress.is.invalid> wrote:

>Steve <lamma...@gmail.com> wrote:

>>>>Date: Fri, 1 Jan 2021 03:21:34 -0800 (PST) <----

>>>>Injection-Date: Fri, 01 Jan 2021 11:21:35 +0000 <----

>>The problem as I see it is Google puts their timezone
>>(PST) into the 'Date' header

> If 'notyalckram' is a UK poster, that's indeed a silly thing to do for
>Google.

No, it's not, for Google Groups doesn't allow the user to state local
time. From the perspective of Google Groups, local time with
respect to Google headquartes is -0800/-0700. Google Groups is just a
Web interface. The client is remote to the user.

>>and adds an 'Injection- Date' header with the posters timezone which
>>Gravity ignores.

> The 'Injection-Date' header is indeed reported in UTC, but that does

>not mean that the poster lives in the UTC timezone. . . .

+0000 is a statement that UTC is used as local time, like GMT as one's
local time zone. -0000 is UTC without reference to local time.

> All in all, I think that as - as I described before - 'Date' and
>'Injection-Date' contain the exact same time, it looks like Gravity can
>not handle the syntactically correct 'Date' header. That's rather
>strange, because using a timezone code like '(PST)' is long standing
>common practice.

It's in parentheses and MUST BE ignored as a comment. As you and others
correctly pointed out, the four digit time zone offset is standard and
that's what Gravity has failed to parse.

PST, not in parentheses, is the time zone offset obsoleted by RFC 2822. RFC
822 had multiple standard ways of expressing the time zone offset and
notoriously misstated single letter military time zones.

>>. . .

[Adam H. Kerman]

I owned up to my error and apologized already. Thanks for accepting my
apology like an all-growed-up boy.

[Steve]

On Sun, 3 Jan 2021 21:51:22 -0000 (UTC), Adam H. Kerman

wrote:
>
> Steve <lamma...@gmail.com> wrote:
> >On Sun, 3 Jan 2021 21:14:17 -0000 (UTC), Adam H. Kerman wrote:
>
> >[...]
> >>Wrong.
>
> >Don't you have anything better to do than being a prize
> >prick, Adam?
>
> I owned up to my error and apologized already. Thanks for accepting my
> apology like an all-growed-up boy.

I accept your apology that you wrote *after* my reply.
You're still a prick.

[wib...@bibble.com.invalid]

On Sun, 3 Jan 2021 21:35:04 -0000, Steve <lamma...@gmail.com> wrote:

>Gravity is no longer maintained

Not true.

See https://www.grc.com/freeupdated.htm

Gibson Research has taken it on.

Last updated 24th June 2020

[Steve]

Thanks, wibble (Jim?).
Yes I tried that one and found it didn't fix what I
wanted it to fix so I went back. There's another fork at
<https://github.com/taviso/mpgravity> that you may know
about, but on reading the changelog it doesn't fix the
bugs that I'd like fixed.

[Grant Taylor]

On 1/3/21 4:06 PM, Steve wrote:
> Yes I tried that one and found it didn't fix what I wanted it to fix
> so I went back.

Have you connected with people in the various GRC forms (private NNTP
newsgroups)?

I've witnessed a few different problems fixed beyond what prompted Steve
to tilt at this windmill. -- Enough so that I think that they are
willing to fix other things. So, they might be willing to fix the
problem that's annoying you.

I'd be happy to try to point you in (what I think is) the proper
direction if you are interested.



--
Grant. . . .
unix || die

[Adam H. Kerman]

Steve <lamma...@gmail.com> wrote:
>On Sun, 3 Jan 2021 21:51:22 -0000 (UTC), Adam H. Kerman wrote:
>>Steve <lamma...@gmail.com> wrote:
>>>On Sun, 3 Jan 2021 21:14:17 -0000 (UTC), Adam H. Kerman wrote:

>>>[...]
>>>>Wrong.

>>>Don't you have anything better to do than being a prize
>>>prick, Adam?

>>I owned up to my error and apologized already. Thanks for accepting my
>>apology like an all-growed-up boy.

>I accept your apology that you wrote *after* my reply.

Not after I'd read it.

>You're still a prick.

Let us know when you've gotten over it.

[Michael Bäuerle]

Frank Slootweg wrote:
> > > Steve <lamma...@gmail.com> wrote:
> > > >

> > > > Example header.
> > > >
> > > > [...]

> > > > Date: Fri, 1 Jan 2021 03:21:34 -0800 (PST) <----
> > > > User-Agent: G2/1.0
> > > > MIME-Version: 1.0

> > > > Message-ID: <3be7b521-45b4-4b71...@googlegroups.com>
> > > > [...]

> > > > Injection-Date: Fri, 01 Jan 2021 11:21:35 +0000 <----
>

> [...]

> The 'Injection-Date' header is indeed reported in UTC, but that does
> not mean that the poster lives in the UTC timezone. My 'Date' header is
> also in UTC ('GMT'), but I live in The Netherlands, which (now, no DST)
> is UTC+01:00.
>
> All in all, I think that as - as I described before - 'Date' and
> 'Injection-Date' contain the exact same time, it looks like Gravity can
> not handle the syntactically correct 'Date' header. That's rather
> strange, because using a timezone code like '(PST)' is long standing
> common practice.
>
> Maybe Michael can see if there's anything wrong with the 'Date' header
> (which I don't see).

No, the syntax of the field looks valid.

Sytax definitions quoted from [1]:
|
| date-time = [ day-of-week "," ] date time [CFWS]
| day-of-week = ([FWS] day-name) / obs-day-of-week
| date = day month year
| time = time-of-day zone
| time-of-day = hour ":" minute [ ":" second ]
| zone = (FWS ( "+" / "-" ) 4DIGIT) / obs-zone

The date with the optional <day-of-week> and the time with the optional
<second> are the variants used by most posting/injecting agents.

Date: Fri, 1 Jan 2021 03:21:34 -0800 (PST)

The comment "(PST)" matches the optional <CFWS> at the end of
<date-time>. Therefore it is allowed at this place but has no
semantical meaning, as Adam has already written.

These two variants for more brevity should be correct syntax too:

Date: 1 Jan 2021 03:21:34 -0800
Date: 1 Jan 2021 03:21 -0800

flnews has used the first one until version 0.12 and it was reported
that this breaks the parser of PhoNews (<day-of-week> is now generated
for better compatibility).

The following variant uses "PST" syntactically as <zone>:

Date: Fri, 1 Jan 2021 03:21:34 PST

"PST" matches <obs-zone> [2]:
|
| [...] PST is semantically equivalent to -0800

RFC 5322 says in [3] that generating obsolete syntax is no longer
allowed. But a conformant parser must still understand it:
|
| [...] Though these syntactic forms MUST NOT be generated according
| to the grammar in section 3, they MUST be accepted and parsed by a
| conformant receiver.

RFC 5536 explicitly specifies that "GMT" must be accepted for
<obs-zone> [4], but this only repeats what is already specified
by the reference to RFC 5322 (for "GMT", "PST" and some others).


My conclusion:
In theory Gravity should understand all variants listed above.
Otherwise this should be reported as a bug.


_____________
[1] <https://tools.ietf.org/html/rfc5322#section-3.3>
[2] <https://tools.ietf.org/html/rfc5322#section-4.3>
[3] <https://tools.ietf.org/html/rfc5322#section-4>
[4] <https://tools.ietf.org/html/rfc5536#section-3.1.1>

[Frank Slootweg]

Adam H. Kerman <a...@chinet.com> wrote:

> Frank Slootweg <th...@ddress.is.invalid> wrote:
> >Steve <lamma...@gmail.com> wrote:
>
> >>>>Date: Fri, 1 Jan 2021 03:21:34 -0800 (PST) <----
> >>>>Injection-Date: Fri, 01 Jan 2021 11:21:35 +0000 <----
>
> >>The problem as I see it is Google puts their timezone
> >>(PST) into the 'Date' header
>
> > If 'notyalckram' is a UK poster, that's indeed a silly thing to do for
> >Google.
>
> No, it's not, for Google Groups doesn't allow the user to state local
> time. From the perspective of Google Groups, local time with
> respect to Google headquartes is -0800/-0700. Google Groups is just a
> Web interface. The client is remote to the user.

I beg to differ. It doesn't matter what Google Groups doesn't allow,
it doesn't matter in which timezone Google's headquarters are and it
doesn't matter on which system the client runs.

Google sends articles to NetNews/Usenet and accepts articles from
same. By doing so, Google has to conform to NetNews/Usenet standards.
On NetNews/Usenet 'Date:' is and always has been the (representation of)
*user* local time [1].

But, the point is moot, because while converting the user's local time
to PST-time and adding '(PST)' is *silly*, it's not *wrong*, because the
converted time ('03:21:34 -0800') *is* the correct user's local time
(11:21:34 UTC) and '(PST)' is - as you mentioned - just a comment (which
should be ignored).

Of course I don't expect Google to care/change, especially because
they've never cared much - if at all - about NetNews/Usenet.

BTW, Google *does* know and use the user's local time, because it says
'+0000' (plus, not minus) - i.e. a time zone at Universal Time - in the
Injection-Date:

> >>>>Injection-Date: Fri, 01 Jan 2021 11:21:35 +0000 <----

[...]

[1]
Netnews Article Format
<https://tools.ietf.org/html/rfc5536#section-3.1.1>

3.1.1. Date

The Date header field is the same as that specified in Sections 3.3
and 3.6.1 of [RFC5322],

Internet Message Format
<https://tools.ietf.org/html/rfc5322#section-3.6.1>

3.6.1. The Origination Date Field

The origination date field consists of the field name "Date" followed
by a date-time specification.

orig-date = "Date:" date-time CRLF

The origination date specifies the date and time at which the creator
of the message indicated that the message was complete and ready to
enter the mail delivery system.

[Frank Slootweg]

[Disclaimer: Yes, I saw the tiff/apology.]

Adam H. Kerman <a...@chinet.com> wrote:

> Steve <lamma...@gmail.com> wrote:
[...]

> >> When responding please give your local time zone (and offset from UTC)
>
> >(UTC+00:00) Dublin, Edinburgh, Lisbon, London
>
> Wrong. Please read RFC 2822. If GMT is local time, then it's -0000.
> Because it will screw up parsing on a Date header, the colon as time
> separator is never used.

FWIW, AFAICT, the '(UTC+00:00) ...' is not what's in any article
header, but the *Windows* 'Time zone' setting of Steve's system. That
setting has that format and content.

[Adam H. Kerman]

Fair enough. I'll apologize for misunderstanding that, too, and we can
have several more rounds of Steve calling me a prize prick.

[Adam H. Kerman]

Frank Slootweg <th...@ddress.is.invalid> wrote:
>Adam H. Kerman <a...@chinet.com> wrote:
>>Frank Slootweg <th...@ddress.is.invalid> wrote:
>>>Steve <lamma...@gmail.com> wrote:

>>>>>>Date: Fri, 1 Jan 2021 03:21:34 -0800 (PST) <----
>>>>>>Injection-Date: Fri, 01 Jan 2021 11:21:35 +0000 <----

>>>>The problem as I see it is Google puts their timezone
>>>>(PST) into the 'Date' header

>>> If 'notyalckram' is a UK poster, that's indeed a silly thing to do for
>>>Google.

>>No, it's not, for Google Groups doesn't allow the user to state local
>>time. From the perspective of Google Groups, local time with
>>respect to Google headquartes is -0800/-0700. Google Groups is just a
>>Web interface. The client is remote to the user.

> I beg to differ. It doesn't matter what Google Groups doesn't allow,
>it doesn't matter in which timezone Google's headquarters are and it
>doesn't matter on which system the client runs.

> Google sends articles to NetNews/Usenet and accepts articles from
>same. By doing so, Google has to conform to NetNews/Usenet standards.
>On NetNews/Usenet 'Date:' is and always has been the (representation of)
>*user* local time [1].

Google Groups is incapable of sending a conventional News article in
plain text. That it doesn't allow the user to set local time is
nonstandard, true, but this is the one crappy thing Google Groups does
that breaks nothing.

[Steve]

On Sun, 3 Jan 2021 17:14:27 -0700, Grant Taylor wrote:
>
> On 1/3/21 4:06 PM, Steve wrote:
> > Yes I tried that one and found it didn't fix what I wanted it to fix
> > so I went back.
>
> Have you connected with people in the various GRC forms (private NNTP
> newsgroups)?

Some time ago I had a look in there, but I didn't post.

> I've witnessed a few different problems fixed beyond what prompted Steve
> to tilt at this windmill. -- Enough so that I think that they are
> willing to fix other things. So, they might be willing to fix the
> problem that's annoying you.
>
> I'd be happy to try to point you in (what I think is) the proper
> direction if you are interested.

That'd be great, thanks.

[Steve]

On Mon, 4 Jan 2021 17:40:02 -0000 (UTC), Adam H. Kerman
wrote:
>

No problem. Next time you act like one I'll be sure to
let you know.

[Grant Taylor]

On 1/4/21 2:37 PM, Steve wrote:
> That'd be great, thanks.

I see discussions about MicroPlanet Gravity in the grc.techtalk and
grc.news.feedback newsgroups on GRC's news server.

See https://www.grc.com/discussions.htm for more information on how to
participate.

TL;DR: Point your news reader to news.grc.com and choose a username &
password. (What is is doesn't really matter as long as you use then
consistently. More details in Newsreader Configuration point 2.)

Let me know if you'd like anything more specific than that.