nnml corruption in move to emacs 21?

nnml corruption in move to emacs 21?

[Todd H.]

Greetings, 

I was recently forced into an upgrade of Emacs for the first time in
a few years.   My last emacs was 20.6. 

Now, I have a hugemongous nnml directory for spam thats over 30,000
messages since I rarely bother to go in there unless I'm hunting for
something specific.

I've just noticed that when I moved my home directory to this new
machine, though, something appears to have gotten corrupted in that
subdirectory so when I access this nnml group, only the messages since
the switchover are viewable.

The older messages show up in the message window are garbled, almost
like I'm looking at something uuencoded or compressed.  And, if I look
at the actual files on the disk now, they look similarly garbled, so
it's not just a display issue. 

So, am I chasing down a file corruption issue, or did something change
in Gnus perhaps that might have modified nnml message store, or is
there a known issue when nnml message counts get over a certain
threshhold, or ??? 

TIA for any insight or advice! 

Best Regards, 
-- 
Todd H.
http://www.toddh.net/

Re: nnml corruption in move to emacs 21?

[Reiner Steib]

On Sat, Mar 19 2005, Todd H. wrote:

> Now, I have a hugemongous nnml directory for spam thats over 30,000
> messages since I rarely bother to go in there unless I'm hunting for
> something specific.
>
> I've just noticed that when I moved my home directory to this new
> machine, though, something appears to have gotten corrupted in that
> subdirectory so when I access this nnml group, only the messages since
> the switchover are viewable.
>
> The older messages show up in the message window are garbled, almost
> like I'm looking at something uuencoded or compressed.  And, if I look
> at the actual files on the disk now, they look similarly garbled, so
> it's not just a display issue. 

I think it's not a Gnus issue then.

> So, am I chasing down a file corruption issue, or did something change
> in Gnus perhaps that might have modified nnml message store, or is
> there a known issue when nnml message counts get over a certain
> threshhold, or ??? 

I'd expect problems if the count exceeds the largest Lisp integer.

,----[ C-h v most-positive-fixnum RET ]
| most-positive-fixnum's value is 268435455
| 
| The largest value that is representable in a Lisp integer.
`----

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/

Re: nnml corruption in move to emacs 21?

[Stefan Monnier]

> I'd expect problems if the count exceeds the largest Lisp integer.
> ,----[ C-h v most-positive-fixnum RET ]
> | most-positive-fixnum's value is 268435455

Note that this value changes between different versions of Emacs.
In Emacs versions before 19.34, the value was typically 8388607, whereas in
versions until 21.4, it's typically 134217728 (and in Emacs-CVS it's as
shown above).  But even that changes on 64bit systems (e.g. on my amd64
system, it says 2305843009213693951).

The values are yet different for XEmacs.


        Stefan

Re: nnml corruption in move to emacs 21?

[Reiner Steib]

On Thu, Mar 31 2005, Stefan Monnier wrote:

>> I'd expect problems if the count exceeds the largest Lisp integer.
>> ,----[ C-h v most-positive-fixnum RET ]
>> | most-positive-fixnum's value is 268435455
>
> Note that this value changes between different versions of Emacs.

Yes, I know.  But all off them should be significantly larger than
30,000 (the number of messages Todd mentioned).

> In Emacs versions before 19.34, the value was typically 8388607, whereas in
> versions until 21.4, it's typically 134217728 (and in Emacs-CVS it's as
> shown above).  But even that changes on 64bit systems (e.g. on my amd64
> system, it says 2305843009213693951).

On my amd64 it is 1152921504606846975.  I guess your built isn't a
plain version from the trunk?

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/