total and auto expiry

total and auto expiry

[Alexander Kotelnikov]

Hello.

While reconfiguring my gnus I was confused with 
gnus-auto-expirable-newsgroups and gnus-total-expirable-newsgroups
descriptions:

From info gnus:
1.
Groups that match the regular expression `gnus-auto-expirable-newsgroups'
will have all articles that you read marked as expirable automatically.
2.
Even more dangerous is the `gnus-total-expirable-newsgroups'
variable.  All groups that match this regexp will have all read
articles put through the expiry process, which means that _all_ old
mail articles in the groups in question will be deleted after a while.


Does this mean that the only difference between actions that are taken
over groups matching 'total' and 'auto' it that the former will be
expired even if expiry-wait for it is 'never?

-- 
Alexander Kotelnikov
Saint-Petersburg, Russia

Re: total and auto expiry

[Kai Großjohann]

Alexander Kotelnikov <sacha@forexware.com> writes:

> Does this mean that the only difference between actions that are taken
> over groups matching 'total' and 'auto' it that the former will be
> expired even if expiry-wait for it is 'never?

No.  I think setting expiry-wait to never means that nothing is
expired.

Normally, you can mark a message as expirable by hitting `E'.

Auto-expire means that Gnus hits `E' for you.  Nothing more, nothing
less.

Total-expire means that messages marked E are expirable, as always,
but *also* messages marked r, R, Y, K and so on are considered
expirable.

Adaptive scoring does not work with auto-expire because it depends on
the different marks r, R, Y, K etc.  But using auto-expire means that
messages will be marked E, instead.

It used to be that total-expire was slower than auto-expire.  I don't
know whether this is still true.

-- 
A preposition is not a good thing to end a sentence with.

Re: total and auto expiry

[Alexander Kotelnikov]

>>>>> On Thu, 13 Mar 2003 18:25:32 +0100
>>>>> "KG" == Kai Großjohann <kai.grossjohann@uni-duisburg.de> wrote:
KG> 
KG> Alexander Kotelnikov <sacha@forexware.com> writes:
>> Does this mean that the only difference between actions that are taken
>> over groups matching 'total' and 'auto' it that the former will be
>> expired even if expiry-wait for it is 'never?
KG> 
KG> No.  I think setting expiry-wait to never means that nothing is
KG> expired.
KG> 
KG> Normally, you can mark a message as expirable by hitting `E'.
KG> 
KG> Auto-expire means that Gnus hits `E' for you.  Nothing more, nothing
KG> less.
KG> 
KG> Total-expire means that messages marked E are expirable, as always,
KG> but *also* messages marked r, R, Y, K and so on are considered
KG> expirable.

Thus the one difference is an article's mark. Its destiny, to be or
not to be expired, is the same, if group is matched with total or auto
variable.

This also means that "total" described as dangerous is not too much
more dangerous than "auto", isn't it?

KG> Adaptive scoring does not work with auto-expire because it depends on
KG> the different marks r, R, Y, K etc.  But using auto-expire means that
KG> messages will be marked E, instead.
KG> 
KG> It used to be that total-expire was slower than auto-expire.  I don't
KG> know whether this is still true.

So it can be advised to use total with adaptive scoring and auto
otherwise. At first approximation they are the same.

-- 
Alexander Kotelnikov
Saint-Petersburg, Russia

Re: total and auto expiry

[David S Goldberg]

>>>>> On Thu, 13 Mar 2003 13:34:16 -0500, Alexander Kotelnikov
>>>>> <sacha@myxomop.com> said:

> This also means that "total" described as dangerous is not too much
> more dangerous than "auto", isn't it?

Well I don't think that we should say that auto-expire is for wimps,
but at least with auto-expire it's very explicitly clear to the user
that an article is marked expireable and that the user must take some
action if s/he doesn't want it to be.  Moreover if using total-expire
and a tick or dormant mark (which are not total-expired) is
inadvertently removed and expiry takes place, the article will be
lost.  That is not an issue for auto-expire.  I therefore think
total-expire is more dangerous than auto-expire.

> So it can be advised to use total with adaptive scoring and auto
> otherwise. At first approximation they are the same.

In spite of being bitten by the problems described I use and prefer
total-expire in most of my groups.  I really like adaptive scoring.
And I've got very good backups.

-- 
Dave Goldberg
david.goldberg6@verizon.net

Re: total and auto expiry

[Kai Großjohann]

Alexander Kotelnikov <sacha@myxomop.com> writes:

>>>>>> On Thu, 13 Mar 2003 18:25:32 +0100
>>>>>> "KG" == Kai Großjohann <kai.grossjohann@uni-duisburg.de> wrote:
> KG> 
> KG> Alexander Kotelnikov <sacha@forexware.com> writes:
>>> Does this mean that the only difference between actions that are taken
>>> over groups matching 'total' and 'auto' it that the former will be
>>> expired even if expiry-wait for it is 'never?
> KG> 
> KG> No.  I think setting expiry-wait to never means that nothing is
> KG> expired.
> KG> 
> KG> Normally, you can mark a message as expirable by hitting `E'.
> KG> 
> KG> Auto-expire means that Gnus hits `E' for you.  Nothing more, nothing
> KG> less.
> KG> 
> KG> Total-expire means that messages marked E are expirable, as always,
> KG> but *also* messages marked r, R, Y, K and so on are considered
> KG> expirable.
>
> Thus the one difference is an article's mark. Its destiny, to be or
> not to be expired, is the same, if group is matched with total or auto
> variable.

Yes.

> This also means that "total" described as dangerous is not too much
> more dangerous than "auto", isn't it?

Before you turned on total-expire, you might have read hundreds of
messages which are suddenly considered expirable.

This is (potentially) dangerous.

> KG> Adaptive scoring does not work with auto-expire because it depends on
> KG> the different marks r, R, Y, K etc.  But using auto-expire means that
> KG> messages will be marked E, instead.
> KG> 
> KG> It used to be that total-expire was slower than auto-expire.  I don't
> KG> know whether this is still true.
>
> So it can be advised to use total with adaptive scoring and auto
> otherwise. At first approximation they are the same.

Except for the above, yes.

Another difference is that total-expire means that you lose one
mark.  With auto-expire, you can distinguish between ticked (!),
dormant (?) and read (R) articles, all of which are kept.  With
total-expire, you only have ticked and dormant.

-- 
A preposition is not a good thing to end a sentence with.