How to run with no `marks' code at all

How to run with no `marks' code at all

[Harry Putnam]

I've had about enough of the problems associated with the new marks
code, or maybe its some other new code, but I see a continuous problem
in one nnml group that I use for a prescan group.  Not the ordinary
usage to be sure but has been working until `marks' made there debut.

The prescan group slurps a procmail spoolfile that contains all new
mail at each fetch.  Once I scan it there, I respool it with my split
rules so that group is always left empty after a respool.

Here lately I find already respooled messages still in there only not
really, the files are gone but .overview or newsrc.eld still sees
them.

Usually have to generate nov after destroying .overview .marks.  I
never saw this kind of thing prior to `marks'

Also it seems that a `g' command now takes quite a lot longer than
before.

So to get to the punch line:  How can I disable all marks related
stuff in order to see if that is the culprit?  Or maybe just retreat
to an earlier gnus.   Not really sure how to checkout a specific
version or which one to try for even.

Re: How to run with no `marks' code at all

[Paul Jarc]

Harry Putnam <reader@newsguy.com> wrote:
> Here lately I find already respooled messages still in there only not
> really, the files are gone but .overview or newsrc.eld still sees
> them.

You might try nnmaildir. :)  No such behavior here, AFAIK.

> How can I disable all marks related stuff in order to see if that is
> the culprit?

Set nnml-marks-is-evil to t in your server parameters.


paul

Re: How to run with no `marks' code at all

[Simon Josefsson]

On Mon, 3 Sep 2001, Harry Putnam wrote:

> The prescan group slurps a procmail spoolfile that contains all new
> mail at each fetch.  Once I scan it there, I respool it with my split
> rules so that group is always left empty after a respool.
>
> Here lately I find already respooled messages still in there only not
> really, the files are gone but .overview or newsrc.eld still sees
> them.

How does your respooling setup look like?  Is the respooling commands
invoked in the prescan summary buffer?  What does *Messages* contain after
you have done all this?

> Also it seems that a `g' command now takes quite a lot longer than
> before.

Yes.  I guess I'll add the obvious speedup now, and track down other
possible optimizations later.  Details about which functions are slow are
helpful here.

> So to get to the punch line:  How can I disable all marks related
> stuff in order to see if that is the culprit?

Put (nnml-marks-is-evil t) as a server parameter.

> Or maybe just retreat to an earlier gnus.   Not really sure how to
> checkout a specific version or which one to try for even.

`cvs update -D FOO' or `cvs update -r BAR' should work, where FOO being a
date (search in the ChangeLog for `marks' or something), BAR being a CVS
tag.