curiosity, question on integer-indices of articles

[Julieta Shem]

In my configuration I think that Gnus v5.13 estimates the number of new
articles by looking at the group-command answer

group comp.programming
211 11994 4454 29913 comp.programming

I suppose it checks which article it has seen last and supposing the
last article it has seen was numbered 29910, it would then conclude that
there are three new articles in comp.programming. When I try to fetch
these possible three new articles, I discover they don't exist.

--8<---------------cut here---------------start------------->8---
article 29913
423 No such article number 29913
article 29912
423 No such article number 29912
article 29911
423 No such article number 29911
--8<---------------cut here---------------end--------------->8---

In the specific case of Eternal September, is this usually a deletion of
articles due to the typical spam?

[Retro Guy]

When an article is deleted, the article number remains, meaning that the
next article added to the spool has the next number. So the deleted
article's number will remain but the article will no longer exist. (I hope
that makes sense :)

Servers that honor NoCeM messages will delete the articles listed in the
NoCeM message, so those article numbers will no longer have corresponding
articles.

With the significant increase in spam over the past few months, there will
be a lot more "article does not exist" messages from servers, and what
appears to be a miscalculation of available articles. (appears to be is
key)

[Julieta Shem]

Retro Guy <retr...@novabbs.org> writes:

> On Sun, 07 Jan 2024 14:13:33 -0300, Julieta Shem wrote:
>
>> In my configuration I think that Gnus v5.13 estimates the number of new
>> articles by looking at the group-command answer
>>
>> group comp.programming
>> 211 11994 4454 29913 comp.programming
>>
>> I suppose it checks which article it has seen last and supposing the
>> last article it has seen was numbered 29910, it would then conclude that
>> there are three new articles in comp.programming. When I try to fetch
>> these possible three new articles, I discover they don't exist.
>>
>> --8<---------------cut here---------------start------------->8---
>> article 29913
>> 423 No such article number 29913
>> article 29912
>> 423 No such article number 29912
>> article 29911
>> 423 No such article number 29911
>> --8<---------------cut here---------------end--------------->8---
>>
>> In the specific case of Eternal September, is this usually a deletion of
>> articles due to the typical spam?
>
> When an article is deleted, the article number remains, meaning that the
> next article added to the spool has the next number. So the deleted
> article's number will remain but the article will no longer exist. (I hope
> that makes sense :)

It does. Thanks!

> Servers that honor NoCeM messages will delete the articles listed in the
> NoCeM message, so those article numbers will no longer have corresponding
> articles.
>
> With the significant increase in spam over the past few months, there will
> be a lot more "article does not exist" messages from servers, and what
> appears to be a miscalculation of available articles. (appears to be is
> key)

Got ya.

[Ray Banana]

Thus spake Julieta Shem <js...@yaxenu.org>

> I suppose it checks which article it has seen last and supposing the
> last article it has seen was numbered 29910, it would then conclude that
> there are three new articles in comp.programming.

That is correct.

> article 29913
> 423 No such article number 29913
> article 29912
> 423 No such article number 29912
> article 29911
> 423 No such article number 29911
>

> In the specific case of Eternal September, is this usually a deletion of
> articles due to the typical spam?

In the current situation these articles were most likely removed by
NoCeM messages issued by other servers. If an article is considered spam
by Eternal-September's filters, the article will be rejected immediately
and hence will not show up in the overview data at all.

--
Пу́тін — хуйло́
https://www.eternal-september.org

[Julieta Shem]

Ray Banana <ray...@raybanana.net> writes:

[...]

>> article 29913
>> 423 No such article number 29913
>> article 29912
>> 423 No such article number 29912
>> article 29911
>> 423 No such article number 29911
>>
>> In the specific case of Eternal September, is this usually a deletion of
>> articles due to the typical spam?
>
> In the current situation these articles were most likely removed by
> NoCeM messages issued by other servers. If an article is considered spam
> by Eternal-September's filters, the article will be rejected immediately
> and hence will not show up in the overview data at all.

Just to confirm --- Eternal-September classification of spam effectively
means the article has never existed, that is, it does not increment the
article number counter at all. Is that what you said? Thanks!

[Ray Banana]

Thus spake Julieta Shem <js...@yaxenu.org>

> Just to confirm --- Eternal-September classification of spam effectively
> means the article has never existed, that is, it does not increment the
> article number counter at all. Is that what you said? Thanks!

Almost. The article for the target group will not be incremented, the
article is not stored in the server spool and overview, but the message
ID is recorded as rejected in the history file. And, of course, the
article will not be propagated to peers.

HTH

[Julieta Shem]

Ray Banana <ray...@raybanana.net> writes:

> Thus spake Julieta Shem <js...@yaxenu.org>
>
>> Just to confirm --- Eternal-September classification of spam effectively
>> means the article has never existed, that is, it does not increment the
>> article number counter at all. Is that what you said? Thanks!
>
> Almost. The article for the target group will not be incremented, the
> article is not stored in the server spool and overview, but the message
> ID is recorded as rejected in the history file. And, of course, the
> article will not be propagated to peers.

Okay, so I can't say it ``never existed'', but it does not increment the
article id counter from the perspective of the overview. Thanks!

[Julieta Shem]

Julieta Shem <js...@yaxenu.org> writes:

> Ray Banana <ray...@raybanana.net> writes:
>
>> Thus spake Julieta Shem <js...@yaxenu.org>
>>
>>> Just to confirm --- Eternal-September classification of spam effectively
>>> means the article has never existed, that is, it does not increment the
>>> article number counter at all. Is that what you said? Thanks!
>>
>> Almost. The article for the target group will not be incremented, the
>> article is not stored in the server spool and overview, but the message
>> ID is recorded as rejected in the history file. And, of course, the
>> article will not be propagated to peers.

I don't know how smart other news readers are, but Gnus seems to figure
out whether there are new posts in a group by using the GROUP command.
If the high-numbered article is greater than the last one Gnus has seen,
it will tell me it has new articles --- but in reality that was just
destroyed spam (hopefully) so I find nothing when I enter the group.

Of course, it has the advantage that I get to discover there was a spam.
In fact, I can say that approximately once a day comp.lang.lisp gets a
few posts that do not make it to the group. I can't see any ideal
solution there.

Just to be clear, I prefer the current behavior.