nntp marks getting reset all the time

nntp marks getting reset all the time

[Ted Zlatanov]

On nntp my marks are getting reset semi-randomly to older values.  Old
`d' read messages show up in the summary buffer again.  It's very
annoying.  I've been coping by saving the marks with gnus-sync, but
don't know what I should be bisecting or debugging because it only
happens occasionally.

Ted

Re: nntp marks getting reset all the time

[Steinar Bang]

>>>>> Ted Zlatanov <tzz@lifelogs.com>:

> On nntp my marks are getting reset semi-randomly to older values.  Old
> `d' read messages show up in the summary buffer again.  It's very
> annoying.  I've been coping by saving the marks with gnus-sync, but
> don't know what I should be bisecting or debugging because it only
> happens occasionally.

The times I've seen this have been early in the phase of starting to
use gnus-sync on a group.  When I have gnus-sync'd for a while things
have been stable, or at least: seemingly stable.

But when I "lost" gnus-sync during the .authinfo changing around last
week, I've lost read marks from back to the time when I first started
gnus-sync'ing. 

The marks that are left are, I think, those that have been set on
exactly that gnus, and residing in that gnus' .newsrc.eld.

I've been stepping through the gnus-sync merge logic, but that didn't
make me any wiser.  What I understood seemed to operate as expected to
me, if I remember correctly (it's been a while).

Re: nntp marks getting reset all the time

[Ted Zlatanov]

On Wed, 23 Feb 2011 14:45:59 +0100 Steinar Bang <sb@dod.no> wrote: 

>>>>>> Ted Zlatanov <tzz@lifelogs.com>:
>> On nntp my marks are getting reset semi-randomly to older values.  Old
>> `d' read messages show up in the summary buffer again.  It's very
>> annoying.  I've been coping by saving the marks with gnus-sync, but
>> don't know what I should be bisecting or debugging because it only
>> happens occasionally.

SB> The times I've seen this have been early in the phase of starting to
SB> use gnus-sync on a group.  When I have gnus-sync'd for a while things
SB> have been stable, or at least: seemingly stable.

SB> But when I "lost" gnus-sync during the .authinfo changing around last
SB> week, I've lost read marks from back to the time when I first started
SB> gnus-sync'ing. 

SB> The marks that are left are, I think, those that have been set on
SB> exactly that gnus, and residing in that gnus' .newsrc.eld.

SB> I've been stepping through the gnus-sync merge logic, but that didn't
SB> make me any wiser.  What I understood seemed to operate as expected to
SB> me, if I remember correctly (it's been a while).

Yeah, this is something screwy with Gnus itself.

Ted

Re: nntp marks getting reset all the time

[Steinar Bang]

>>>>> Ted Zlatanov <tzz@lifelogs.com>:

> Yeah, this is something screwy with Gnus itself.

Now I'm seeing the messages I read yesterday as unread, on a different,
but gnus-sync'd, gnus.  But after a couple of days of this, it will
probably (hopefully...) stabilize again, and things will be sync'ing
correctly.

It's similar in behaviour to the M-g/QRESYNC thing with nnimap.
Probably not related in cause, though...?

Re: nntp marks getting reset all the time

[Ted Zlatanov]

On Thu, 24 Feb 2011 08:37:12 +0100 Steinar Bang <sb@dod.no> wrote: 

>>>>>> Ted Zlatanov <tzz@lifelogs.com>:
>> Yeah, this is something screwy with Gnus itself.

SB> Now I'm seeing the messages I read yesterday as unread, on a different,
SB> but gnus-sync'd, gnus.  But after a couple of days of this, it will
SB> probably (hopefully...) stabilize again, and things will be sync'ing
SB> correctly.

SB> It's similar in behaviour to the M-g/QRESYNC thing with nnimap.
SB> Probably not related in cause, though...?

Patience...  I'll get to the sync stuff ASAP.

Ted

Re: nntp marks getting reset all the time

[Steinar Bang]

>>>>> Ted Zlatanov <tzz@lifelogs.com>:

> Patience...  I'll get to the sync stuff ASAP.

Ah, you'll work more on it...? :-)

(I was't complaining.  Just supplying data points.  As you said there's
probably something weird in the gnus code itself)

cloudy Gnus (was: nntp marks getting reset all the time)

[Ted Zlatanov]

On Thu, 24 Feb 2011 18:34:56 +0100 Steinar Bang <sb@dod.no> wrote: 

>>>>>> Ted Zlatanov <tzz@lifelogs.com>:
>> Patience...  I'll get to the sync stuff ASAP.

SB> Ah, you'll work more on it...? :-)

SB> (I was't complaining.  Just supplying data points.  As you said there's
SB> probably something weird in the gnus code itself)

Yeah, but gnus-sync it probably triggering the bugs.  Cloudy Gnus is on
my TODO list after someday finishing auth-source.el and bringing GnuTLS
up to 2.1x in Emacs.  I don't know how much time Lars will have for it
but it's definitely time to give Gnus that functionality so I'll do what
I can.  I actually thought I'd be done with auth-source by now but the
damn UI is too complicated.  UI is hard, let's go shopping!

Ted

Re: nntp marks getting reset all the time

[Lars Ingebrigtsen]

Ted Zlatanov <tzz@lifelogs.com> writes:

> On nntp my marks are getting reset semi-randomly to older values.  Old
> `d' read messages show up in the summary buffer again.  It's very
> annoying.  I've been coping by saving the marks with gnus-sync, but
> don't know what I should be bisecting or debugging because it only
> happens occasionally.

Perhaps there's some interaction between NNTP marks and gnus-sync
somewhere, and the NNTP marks are incomplete?

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: nntp marks getting reset all the time

[Ted Zlatanov]

On Thu, 24 Feb 2011 19:01:20 -0800 Lars Ingebrigtsen <larsi@gnus.org> wrote: 

LI> Ted Zlatanov <tzz@lifelogs.com> writes:
>> On nntp my marks are getting reset semi-randomly to older values.  Old
>> `d' read messages show up in the summary buffer again.  It's very
>> annoying.  I've been coping by saving the marks with gnus-sync, but
>> don't know what I should be bisecting or debugging because it only
>> happens occasionally.

LI> Perhaps there's some interaction between NNTP marks and gnus-sync
LI> somewhere, and the NNTP marks are incomplete?

Yeah.  But it stopped again (only happening on nnmaildir now).  I'll
tackle this with the Cloudy Gnus work.

Ted

Re: cloudy Gnus

[Ted Zlatanov]

On Thu, 24 Feb 2011 11:47:04 -0600 Ted Zlatanov <tzz@lifelogs.com> wrote: 

TZ> Yeah, but gnus-sync it probably triggering the bugs.  Cloudy Gnus is on
TZ> my TODO list after someday finishing auth-source.el and bringing GnuTLS
TZ> up to 2.1x in Emacs.  I don't know how much time Lars will have for it
TZ> but it's definitely time to give Gnus that functionality so I'll do what
TZ> I can.  I actually thought I'd be done with auth-source by now but the
TZ> damn UI is too complicated.  UI is hard, let's go shopping!

I disabled the automatic `gnus-sync-read' calls in the Gnus hooks, so
now loading gnus-sync.el and enabling it is harmless.  You'll just get a
file saved if you set the `gnus-sync-backend'.  You have to call
`gnus-sync-read' yourself when you want to load that data.

I wrote in the comments that this is a temporary measure.  I'd like to
get started on Cloudy Gnus soon.  Can we close the book on No Gnus?  Or
is there something yet to be done?

Thanks
Ted

Re: cloudy Gnus

[Lars Magne Ingebrigtsen]

Ted Zlatanov <tzz@lifelogs.com> writes:

> I wrote in the comments that this is a temporary measure.  I'd like to
> get started on Cloudy Gnus soon.  Can we close the book on No Gnus?  Or
> is there something yet to be done?

I think there are still some issues with nnimap that should be sorted
out.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: cloudy Gnus

[Ted Zlatanov]

On Tue, 15 Mar 2011 16:54:07 +0100 Lars Magne Ingebrigtsen <larsi@gnus.org> wrote: 

LMI> Ted Zlatanov <tzz@lifelogs.com> writes:
>> I wrote in the comments that this is a temporary measure.  I'd like to
>> get started on Cloudy Gnus soon.  Can we close the book on No Gnus?  Or
>> is there something yet to be done?

LMI> I think there are still some issues with nnimap that should be sorted
LMI> out.

The asynchronous stuff?  Or something else?  If you can make a list of
release-blocking No Gnus issues and bugs it would be great.

Ted

Re: cloudy Gnus

[Lars Magne Ingebrigtsen]

Ted Zlatanov <tzz@lifelogs.com> writes:

> The asynchronous stuff?  Or something else?  If you can make a list of
> release-blocking No Gnus issues and bugs it would be great.

The release-blocking issue is really that nnimap still seems to be
losing marks for some people.  I've been using nnimap as my primary mail
backend for many months now trying to see whether it happens to me, but
so far it hasn't.

So it's one of those annoying subtle bugs.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: cloudy Gnus

[Ted Zlatanov]

On Tue, 15 Mar 2011 17:11:27 +0100 Lars Magne Ingebrigtsen <larsi@gnus.org> wrote: 

LMI> Ted Zlatanov <tzz@lifelogs.com> writes:
>> The asynchronous stuff?  Or something else?  If you can make a list of
>> release-blocking No Gnus issues and bugs it would be great.

LMI> The release-blocking issue is really that nnimap still seems to be
LMI> losing marks for some people.  I've been using nnimap as my primary mail
LMI> backend for many months now trying to see whether it happens to me, but
LMI> so far it hasn't.

LMI> So it's one of those annoying subtle bugs.

If it's just one issue I think it's reasonable to go ahead with the
release.  But if you want to fix it first, can you collect the bug
information in one place so I can try to find it?  The other Gnus
developers may be interested in a restatement of the bug as well.

Ted

Re: cloudy Gnus

[Lars Magne Ingebrigtsen]

Ted Zlatanov <tzz@lifelogs.com> writes:

> But if you want to fix it first, can you collect the bug information
> in one place so I can try to find it?  The other Gnus developers may
> be interested in a restatement of the bug as well.

It's on several different threads, and there may be different bugs
involved.  The most recent one was the one about the O-marked articles.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: cloudy Gnus

[Didier Verna]

Ted Zlatanov <tzz@lifelogs.com> wrote:

> I'd like to get started on Cloudy Gnus soon.

  Wow, wow, slow down. [...] then Pterodactyl, then No, then Cloudy? So
what happens to the M. series ? I was soooooo expecting Mitocondria Gnus
to fork off...

-- 
Resistance is futile. You will be jazzimilated.

Scientific site:   http://www.lrde.epita.fr/~didier
Music (Jazz) site: http://www.didierverna.com

Re: cloudy Gnus

[Lars Magne Ingebrigtsen]

Didier Verna <didier@xemacs.org> writes:

>   Wow, wow, slow down. [...] then Pterodactyl, then No, then Cloudy? So
> what happens to the M. series ? I was soooooo expecting Mitocondria Gnus
> to fork off...

The next one will be an M series, but we need to find a Japanese word
that starts with M and has some nice kanji. 

That's also a release-blocker.  :-)

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: cloudy Gnus

[Adam Sjøgren]

On Tue, 15 Mar 2011 18:02:22 +0100, Lars wrote:

> The next one will be an M series, but we need to find a Japanese word
> that starts with M and has some nice kanji. 

Also, t-shirts, mugs or somesuch? I think you already did the design?


   :-),

     Adam

-- 
 "Money always takes the place of life"                       Adam Sjøgren
                                                         asjo@koldfront.dk

Re: cloudy Gnus

[Ted Zlatanov]

On Tue, 15 Mar 2011 17:59:58 +0100 Didier Verna <didier@xemacs.org> wrote: 

dvl> Ted Zlatanov <tzz@lifelogs.com> wrote:
>> I'd like to get started on Cloudy Gnus soon.

dvl>   Wow, wow, slow down. [...] then Pterodactyl, then No, then Cloudy? So
dvl> what happens to the M. series ? I was soooooo expecting Mitocondria Gnus
dvl> to fork off...

I suggested Mammatus and Mediocris (types of clouds).  Lars wanted a
Japanese name and unfortunately "Kumo" is Japanese for cloud.  People, I
give you *comedy gold* here!  Gold!

Why not skip to K?  Do we have to go through L and M?  I mean, it's not
like the letters L and M matter to anyone, right Lars Magne?

Ted

Re: cloudy Gnus

[Lars Magne Ingebrigtsen]

asjo@koldfront.dk (Adam Sjøgren) writes:

> Also, t-shirts, mugs or somesuch? I think you already did the design?

I have a design in mind, but I haven't actually finished it...

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: cloudy Gnus

[Lars Magne Ingebrigtsen]

Ted Zlatanov <tzz@lifelogs.com> writes:

> Why not skip to K?  Do we have to go through L and M?  I mean, it's not
> like the letters L and M matter to anyone, right Lars Magne?

*gasp*

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: cloudy Gnus

[Didier Verna]

Lars Magne Ingebrigtsen wrote:

> Didier Verna <didier@xemacs.org> writes:
>
>>   Wow, wow, slow down. [...] then Pterodactyl, then No, then Cloudy? So
>> what happens to the M. series ? I was soooooo expecting Mitocondria Gnus
>> to fork off...
>
> The next one will be an M series, but we need to find a Japanese word
> that starts with M and has some nice kanji. 
>
> That's also a release-blocker.  :-)

  Some suggestions:

- Meiji Gnus (明治) (historical period where the capital moved from
  Kyoto to Tokyo)

- Maguro Gnus (鮪) (means tuna; a single character, funnier I think)

-- 
Resistance is futile. You will be jazzimilated.

Scientific site:   http://www.lrde.epita.fr/~didier
Music (Jazz) site: http://www.didierverna.com

Re: cloudy Gnus

[Ted Zlatanov]

On Wed, 16 Mar 2011 09:15:32 +0100 Didier Verna <didier@xemacs.org> wrote: 

dvl>   Some suggestions:

dvl> - Meiji Gnus (明治) (historical period where the capital moved from
dvl>   Kyoto to Tokyo)

dvl> - Maguro Gnus (鮪) (means tuna; a single character, funnier I think)

I like Maguro, it keeps with the animal theme and they move in large
quantities (connecting with the "cloudy" idea).  Plus they make
wonderful sushi, unlike pterodactyls.

Ted

Re: cloudy Gnus

[Lars Magne Ingebrigtsen]

Ted Zlatanov <tzz@lifelogs.com> writes:

> I like Maguro, it keeps with the animal theme and they move in large
> quantities (connecting with the "cloudy" idea).  Plus they make
> wonderful sushi, unlike pterodactyls.

Something more obscure, i.e., something you don't find on a menu, would
be nice.  :-)

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: cloudy Gnus

[Frank Schmitt]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Ted Zlatanov <tzz@lifelogs.com> writes:
>
>> I like Maguro, it keeps with the animal theme and they move in large
>> quantities (connecting with the "cloudy" idea).  Plus they make
>> wonderful sushi, unlike pterodactyls.
>
> Something more obscure, i.e., something you don't find on a menu, would
> be nice.  :-)

More Gnus?

-- 
Have you ever considered how much text can fit in eighty columns?  Given that a
signature typically contains up to four lines of text, this space allows you to
attach a tremendous amount of valuable information to your messages.  Seize the
opportunity and don't waste your signature on bullshit that nobody cares about.

Re: cloudy Gnus

[Steinar Bang]

>>>>> Ted Zlatanov <tzz@lifelogs.com>:

> On Tue, 15 Mar 2011 16:54:07 +0100 Lars Magne Ingebrigtsen <larsi@gnus.org> wrote: 

LMI> I think there are still some issues with nnimap that should be sorted
LMI> out.

> The asynchronous stuff?  Or something else?

nnimap and agent aren't all there yet.  Strange things happen (articles
aren't cached in the agent, sometimes the wrong article is cached, and
sometimes the cached "article" is actually filled with NOV lines).

Re: cloudy Gnus

[Didier Verna]

Lars Magne Ingebrigtsen wrote:

> Ted Zlatanov <tzz@lifelogs.com> writes:
>
>> I like Maguro, it keeps with the animal theme and they move in large
>> quantities (connecting with the "cloudy" idea).  Plus they make
>> wonderful sushi, unlike pterodactyls.
>
> Something more obscure, i.e., something you don't find on a menu, would
> be nice.  :-)

Mekajiki Gnus (目梶木): means swordfish. You /can/ find that in a menu,
but much, much less often... still a fish, animal theme etc. ;-)

-- 
Resistance is futile. You will be jazzimilated.

Scientific site:   http://www.lrde.epita.fr/~didier
Music (Jazz) site: http://www.didierverna.com

Re: cloudy Gnus

[Steinar Bang]

>>>>> Ted Zlatanov <tzz@lifelogs.com>:

> On Thu, 24 Feb 2011 18:34:56 +0100 Steinar Bang <sb@dod.no> wrote: 
>>>>>>> Ted Zlatanov <tzz@lifelogs.com>:
>>> Patience...  I'll get to the sync stuff ASAP.

SB> Ah, you'll work more on it...? :-)

SB> (I was't complaining.  Just supplying data points.  As you said there's
SB> probably something weird in the gnus code itself)

> Yeah, but gnus-sync it probably triggering the bugs.

It has been working for a while now, but since sometime last week,
gnus-sync of nntp read marks across my 3 gnusen, has stopped working. 

(I have no idea why it started working again, or why it has stopped
working... also again.  Just supplying a data point)

Re: cloudy Gnus

[Ted Zlatanov]

On Fri, 25 Mar 2011 09:17:24 +0100 Steinar Bang <sb@dod.no> wrote: 

>>>>>> Ted Zlatanov <tzz@lifelogs.com>:
>> On Thu, 24 Feb 2011 18:34:56 +0100 Steinar Bang <sb@dod.no> wrote: 
>>>>>>>> Ted Zlatanov <tzz@lifelogs.com>:
>>>> Patience...  I'll get to the sync stuff ASAP.

SB> Ah, you'll work more on it...? :-)

SB> (I was't complaining.  Just supplying data points.  As you said there's
SB> probably something weird in the gnus code itself)

>> Yeah, but gnus-sync it probably triggering the bugs.

SB> It has been working for a while now, but since sometime last week,
SB> gnus-sync of nntp read marks across my 3 gnusen, has stopped working. 

SB> (I have no idea why it started working again, or why it has stopped
SB> working... also again.  Just supplying a data point)

I would check the mark sync file.  It's probably not being written
correctly.

Ted

Re: cloudy Gnus

[Steinar Bang]

>>>>> Ted Zlatanov <tzz@lifelogs.com>:

> I would check the mark sync file.  It's probably not being written
> correctly.

What should I check for.  The write date looks ok, and the content looks
ok as well.

Also, this gnus, which hadn't been updated for a while, synced
correctly.

Unfortunately I just pulled from the origin repo...:-/

Is there a way to find out what commit I was at before I pulled?  It has
to be some change between there and now... 

All .elc files have the time stamp 2011-03-08 18:37.

Hm... that means the problem was introduced at least after this one:
 commit 2c474a0772b01d6081354487d2d5680d830b0714
 Merge: 0b7a05e 4c756ae
 Author: Ted Zlatanov <tzz@lifelogs.com>
 Date:   Sun Mar 6 05:00:38 2011 -0600

    Merge branch 'master' of https://git.gnus.org/gnus

Re: cloudy Gnus

[Bernt Hansen]

Steinar Bang <sb@dod.no> writes:

>>>>>> Ted Zlatanov <tzz@lifelogs.com>:
>
>> I would check the mark sync file.  It's probably not being written
>> correctly.
>
> What should I check for.  The write date looks ok, and the content looks
> ok as well.
>
> Also, this gnus, which hadn't been updated for a while, synced
> correctly.
>
> Unfortunately I just pulled from the origin repo...:-/
>
> Is there a way to find out what commit I was at before I pulled?  It has
> to be some change between there and now... 

Hi Steinar,

If you just pulled then a temporary ref ORIG_HEAD points at your
previous commit.  You can also use the reflog to get back to recent
points in time in your working repository.

HTH,
-- 
Bernt

Re: cloudy Gnus

[Steinar Bang]

>>>>> Bernt Hansen <bernt@norang.ca>:

> If you just pulled then a temporary ref ORIG_HEAD points at your
> previous commit.  You can also use the reflog to get back to recent
> points in time in your working repository.

Ah, thanks.  The .git/ORIG_HEAD contains 1a2936cbc8b4b584becd8426ac4959957e0595fd
ie. this one
 http://git.gnus.org/cgit/gnus.git/commit/?id=1a2936cbc8b4b584becd8426ac4959957e0595fd

Now I just need to find the energy to do a bisecting thing to see if I
can find what commit changed the behaviour...:-)

Re: cloudy Gnus

[Steinar Bang]

>>>>> Steinar Bang <sb@dod.no>:

> Ah, thanks.  The .git/ORIG_HEAD contains 1a2936cbc8b4b584becd8426ac4959957e0595fd
> ie. this one
>  http://git.gnus.org/cgit/gnus.git/commit/?id=1a2936cbc8b4b584becd8426ac4959957e0595fd

> Now I just need to find the energy to do a bisecting thing to see if I
> can find what commit changed the behaviour...:-)

Ok, done bisecting:

sb@aa1steinar:~/git/gnus$ git bisect good
f48235eacef5c56ad94704a17e438f9f6039ab71 is the first bad commit
commit f48235eacef5c56ad94704a17e438f9f6039ab71
Author: Ted Zlatanov <tzz@lifelogs.com>
Date:   Fri Mar 11 13:39:30 2011 -0600

    Adjust documentation and remove automatic read in hopes for Cloudy Gnus.
    
    * gnus-sync.el (gnus-sync-install-hooks, gnus-sync-unload-hook): Don't
    install `gnus-sync-read' to any hooks by default.  It's buggy.  The
    user will have to run `gnus-sync-read' manually and wait for Cloudy
    Gnus.

:040000 040000 6754ee1e53a386eb40a0c040b6c9ea79842b8f62 afd0a0c9b8e401ac728aadeff00140afa9645b49 M	lisp


Ah... that explains it, I guess...?