loosing read marks in nnml:mail.misc

loosing read marks in nnml:mail.misc

[Alex Schroeder]

I reported this as an inexplicable bug where read marks were getting
lost.  I then found out that my .marks file contains weirdo data.  It
happened again, and I found weirdo data in my .marks file again.
Here are the read marks:

(read (1 . 4975) (5020 . 5023) (5025 . 5018) (5020 . 5013) (5015 .
5011) (5013 . 5008) (5010 . 5005) (5007 . 4997) (4999 . 4975) (4977 .
4974) (4976 . 5031))

Note the cons cells where the start is higher than the end.  These
seem to confuse Gnus to no end.  Furthermore note how there are two
cons cells that would cover all the others: The first one and the last
one.

Does anybody know how the bogus cons cells could creep in, or does
anybody know of something I might want to do to debug this?

Alex.
-- 
http://www.emacswiki.org/alex/
There is no substitute for experience.

Re: loosing read marks in nnml:mail.misc

[Gaute B Strokkenes]

ON 26 aug 2003, alex@emacswiki.org wrote:

> I reported this as an inexplicable bug where read marks were getting
> lost.  I then found out that my .marks file contains weirdo data.
> It happened again, and I found weirdo data in my .marks file again.
> Here are the read marks:
>
> (read (1 . 4975) (5020 . 5023) (5025 . 5018) (5020 . 5013) (5015 .
> 5011) (5013 . 5008) (5010 . 5005) (5007 . 4997) (4999 . 4975) (4977
> .  4974) (4976 . 5031))
>
> Note the cons cells where the start is higher than the end.  These
> seem to confuse Gnus to no end.  Furthermore note how there are two
> cons cells that would cover all the others: The first one and the
> last one.
>
> Does anybody know how the bogus cons cells could creep in, or does
> anybody know of something I might want to do to debug this?

I've had similar things happen to me as well.  So you're not alone.
But I don't know how it could happen either.

-- 
Gaute Strokkenes                        http://www.srcf.ucam.org/~gs234/
Yow!  Am I having fun yet?

Re: loosing read marks in nnml:mail.misc

[Simon Josefsson]

Alex Schroeder <alex@emacswiki.org> writes:

> I reported this as an inexplicable bug where read marks were getting
> lost.  I then found out that my .marks file contains weirdo data.  It
> happened again, and I found weirdo data in my .marks file again.
> Here are the read marks:
>
> (read (1 . 4975) (5020 . 5023) (5025 . 5018) (5020 . 5013) (5015 .
> 5011) (5013 . 5008) (5010 . 5005) (5007 . 4997) (4999 . 4975) (4977 .
> 4974) (4976 . 5031))
>
> Note the cons cells where the start is higher than the end.  These
> seem to confuse Gnus to no end.  Furthermore note how there are two
> cons cells that would cover all the others: The first one and the last
> one.
>
> Does anybody know how the bogus cons cells could creep in, or does
> anybody know of something I might want to do to debug this?

The best would be if you can find some sequence of operations that
introduce the garbage.  Tracing calls to the gnus range functions
could be one way to see if maybe the problem is in them, but checking
if their result is correct might be difficult.

Re: losing read marks in nnml:mail.misc

[Joseph Barillari]

>>>>> "SJ" == Simon Josefsson <jas@extundo.com> writes:

    SJ> Alex Schroeder <alex@emacswiki.org> writes:
    >> I reported this as an inexplicable bug where read marks were
    >> getting lost.  I then found out that my .marks file contains
    >> weirdo data.  It happened again, and I found weirdo data in my
    >> .marks file again.  Here are the read marks:
    >> 
    >> (read (1 . 4975) (5020 . 5023) (5025 . 5018) (5020 . 5013)
    >> (5015 .  5011) (5013 . 5008) (5010 . 5005) (5007 . 4997) (4999
    >> . 4975) (4977 .  4974) (4976 . 5031))
    >> 
    >> Note the cons cells where the start is higher than the end.
    >> These seem to confuse Gnus to no end.  Furthermore note how
    >> there are two cons cells that would cover all the others: The
    >> first one and the last one.
    >> 
    >> Does anybody know how the bogus cons cells could creep in, or
    >> does anybody know of something I might want to do to debug
    >> this?

    SJ> The best would be if you can find some sequence of operations
    SJ> that introduce the garbage.  Tracing calls to the gnus range
    SJ> functions could be one way to see if maybe the problem is in
    SJ> them, but checking if their result is correct might be
    SJ> difficult.

I've got a similar problem, except that gnus now trashes both the
.overview and .marks files: the .overview is truncated to two
messages:

10727   Re: Mail        Joseph Barillari <jbarilla@princeton.edu>       Fri, 19 Sep 2003 11:09:26 -0400 <87y8wkstll.fsf@bigbox.barillari.org>   <87u1gk8y8s.fsf@bigbox.barillari.org> <005701c2b797$a3db5400$0d8efea9@nonamejec54>      310    10       Xref: bigbox.barillari.org mail.mail:10727      To: "Name Elided" <name-elided@foo.com>     

and .marks loses all of the ticked and replied marks:

((forward 98 107 314 407 600 604 669 770 936) (read (1 . 10727)))

I've deleted the .overview so gnus could regenerate it, and replaced
the .marks file with a backup. I've even removed the conses from the
entries in the tick section of .marks. The result is invariably the
same: the general mail group works fine for a while (even letting me
untick articles), and then, all of a sudden, the group shows up
near-empty (thanks to the trashed .overview).

I eventually threw up my hands, entered the server buffer with '^', and
regenerated all of the nnml information with 'g'. I replaced the old
marks file (with the conses removed from the ticks section, though
that was probably unnecessary (as the conses didn't go backwards, as
did Alex Schroeder's). 

So far, the mailbox hasn't gotten hosed again. (Blast, this report
will probably jinx it...)

--Joe

-- 
Joseph Barillari -- http://barillari.org

Re: loosing read marks in nnml:mail.misc

[Lars Magne Ingebrigtsen]

Alex Schroeder <alex@emacswiki.org> writes:

> (read (1 . 4975) (5020 . 5023) (5025 . 5018) (5020 . 5013) (5015 .
> 5011) (5013 . 5008) (5010 . 5005) (5007 . 4997) (4999 . 4975) (4977 .
> 4974) (4976 . 5031))
>
> Note the cons cells where the start is higher than the end.

Huh.  That kinds looks like a general bug in the Gnus range
functions.  On the other hand, that seems kinda unlikely, since you'd
see that all over the place...

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