deleting articles from search groups?

deleting articles from search groups?

[Andrew Cohen]

I just added the ability to delete articles directly from an nnir buffer
(similar to moving them). nnir just passes the delete request up the
food chain to the original group holding the article.

But after doing this, I'm wondering whether or not this is a good
idea. Should we worry that someone might be surprised that an action on
the results of a search has the effect of deleting an article
permanently from its original group? But you might say the same thing
about moving articles. 

A secondary question (assuming that allowing deletes is the Right Thing)
is whether or not I've implemented it correctly:) Deleting articles is
done through the 'request-expire-articles function, so I've created one
for nnir. But this means that gnus-summary-expire-articles is called on
the nnir group upon summary-buffer exit. Stepping through this when
exiting an nnir group always shows the list of "expirable" articles as
nil and so no harm is done. Is this guaranteed? I'm guessing that since
nnir is an "ephemeral" group and doesn't really have marks that the list
of expirable articles (gnus-newsgroup-expirable) is always empty?

Andy

Re: deleting articles from search groups?

[Dan Christensen]

Andrew Cohen <cohen@andy.bu.edu> writes:

> But after doing this, I'm wondering whether or not this is a good
> idea. Should we worry that someone might be surprised that an action on
> the results of a search has the effect of deleting an article
> permanently from its original group? 

Maybe it's worth asking the user, since deletion is permanent?  This is
done in gnus-summary-delete-article if gnus-novice-user is set.  Since
that's the only occurrence of that variable in gnus-sum.el, maybe you
could just set that variable in nnir groups??  Or maybe you want a
customized question:  "Deleting articles will delete them from the
original groups.  Proceed?"

> A secondary question (assuming that allowing deletes is the Right Thing)
> is whether or not I've implemented it correctly:) 

I'll let others answer this part.

Dan

Re: deleting articles from search groups?

[Sivaram Neelakantan]

On Mon, Nov 29 2010,Dan Christensen wrote:

> Andrew Cohen <cohen@andy.bu.edu> writes:
>
>> But after doing this, I'm wondering whether or not this is a good
>> idea. Should we worry that someone might be surprised that an action on
>> the results of a search has the effect of deleting an article
>> permanently from its original group? 
>
> Maybe it's worth asking the user, since deletion is permanent?  This is
> done in gnus-summary-delete-article if gnus-novice-user is set.  Since
> that's the only occurrence of that variable in gnus-sum.el, maybe you
> could just set that variable in nnir groups??  Or maybe you want a
> customized question:  "Deleting articles will delete them from the
> original groups.  Proceed?"
>

I second this.  Though, I think there might be use cases where people
might want to only delete in the searched articles to keep the search
list sane or refine it further by simply deleting the matches that
were false positives.  That's within the search context and not
permanent deletion.



[snipped 10 lines]


 sivaram
 -- 

Re: deleting articles from search groups?

[Rupert Swarbrick]

[-- Attachment #1: Type: text/plain, Size: 872 bytes --]

Sivaram Neelakantan <nsivaram.net@gmail.com> writes:
> On Mon, Nov 29 2010,Dan Christensen wrote:
>> Maybe it's worth asking the user, since deletion is permanent?  This is
>> done in gnus-summary-delete-article if gnus-novice-user is set.  Since
>> that's the only occurrence of that variable in gnus-sum.el, maybe you
>> could just set that variable in nnir groups??  Or maybe you want a
>> customized question:  "Deleting articles will delete them from the
>> original groups.  Proceed?"
>>
>
> I second this.  Though, I think there might be use cases where people
> might want to only delete in the searched articles to keep the search
> list sane or refine it further by simply deleting the matches that
> were false positives.  That's within the search context and not
> permanent deletion.
>

Maybe binding C-k or the like to do this here would make sense?

Rupert

[-- Attachment #2: Type: application/pgp-signature, Size: 315 bytes --]

Re: deleting articles from search groups?

[Sivaram Neelakantan]

On Mon, Nov 29 2010,Rupert Swarbrick wrote:

> Sivaram Neelakantan <nsivaram.net@gmail.com> writes:

[snipped 16 lines]

> Maybe binding C-k or the like to do this here would make sense?
>

Right, that appears logical, using C-k.


 sivaram
 -- 

Re: deleting articles from search groups?

[Lars Magne Ingebrigtsen]

Andrew Cohen <cohen@andy.bu.edu> writes:

> But after doing this, I'm wondering whether or not this is a good
> idea. Should we worry that someone might be surprised that an action on
> the results of a search has the effect of deleting an article
> permanently from its original group? But you might say the same thing
> about moving articles. 

Yes, I don't think users will be surprised that deleting an article in
the nnir group makes it go away forever.  I mean, that's what they're
queried for on `B DEL'.

> But this means that gnus-summary-expire-articles is called on
> the nnir group upon summary-buffer exit.

But this is only called with the "E"-marked articles.  I guess if the
user sets the "E" mark on nnir articles, they probably know that it's in
danger of being deleted.  But I don't think anybody will do that,
really...

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

Re: deleting articles from search groups?

[Andrew Cohen]

>>>>> "Lars" == Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

    Lars> Andrew Cohen <cohen@andy.bu.edu> writes:
    >> But after doing this, I'm wondering whether or not this is a good
    >> idea. Should we worry that someone might be surprised that an
    >> action on the results of a search has the effect of deleting an
    >> article permanently from its original group? But you might say
    >> the same thing about moving articles.

    Lars> Yes, I don't think users will be surprised that deleting an
    Lars> article in the nnir group makes it go away forever.  I mean,
    Lars> that's what they're queried for on `B DEL'.

    >> But this means that gnus-summary-expire-articles is called on the
    >> nnir group upon summary-buffer exit.

    Lars> But this is only called with the "E"-marked articles.  I guess
    Lars> if the user sets the "E" mark on nnir articles, they probably
    Lars> know that it's in danger of being deleted.  But I don't think
    Lars> anybody will do that, really...

OK, I've been living with this for a week and haven't had anything
really bad happen. So I've pushed this. Let me know if anything goes
wrong.

Andy

Re: deleting articles from search groups?

[Dan Christensen]

Andrew Cohen <cohen@andy.bu.edu> writes:

>>>>>> "Lars" == Lars Magne Ingebrigtsen <larsi@gnus.org> writes:
>
>     >> But this means that gnus-summary-expire-articles is called on the
>     >> nnir group upon summary-buffer exit.
>
>     Lars> But this is only called with the "E"-marked articles.  I guess
>     Lars> if the user sets the "E" mark on nnir articles, they probably
>     Lars> know that it's in danger of being deleted.  But I don't think
>     Lars> anybody will do that, really...
>
> OK, I've been living with this for a week and haven't had anything
> really bad happen. So I've pushed this. Let me know if anything goes
> wrong.

It just occurred to me that some people use group parameters to have
different expiry-wait periods for different groups.  Will this above
cause articles to expire based on the expiry-wait for the original group
or whatever expiry-wait applies to the nnir group (probably the default
global value)?

Dan

Re: deleting articles from search groups?

[Lars Magne Ingebrigtsen]

Dan Christensen <jdc@uwo.ca> writes:

> It just occurred to me that some people use group parameters to have
> different expiry-wait periods for different groups.  Will this above
> cause articles to expire based on the expiry-wait for the original group
> or whatever expiry-wait applies to the nnir group (probably the default
> global value)?

The latter, I think.  But the nnir groups are ephemeral, aren't they?
So they shouldn't affect anything.  Unless I'm misunderstanding everything.

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

Re: deleting articles from search groups?

[Andrew Cohen]

>>>>> "Lars" == Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

    Lars> Dan Christensen <jdc@uwo.ca> writes:
    >> It just occurred to me that some people use group parameters to
    >> have different expiry-wait periods for different groups.  Will
    >> this above cause articles to expire based on the expiry-wait for
    >> the original group or whatever expiry-wait applies to the nnir
    >> group (probably the default global value)?

    Lars> The latter, I think.  But the nnir groups are ephemeral,
    Lars> aren't they?  So they shouldn't affect anything.  Unless I'm
    Lars> misunderstanding everything.

My message crossed yours. This is correct: ephemeral, so it shouldn't
cause any trouble (and hasn't in my one week of minimal testing).

Re: deleting articles from search groups?

[Dan Christensen]

Andrew Cohen <cohen@andy.bu.edu> writes:

>>>>>> "Lars" == Lars Magne Ingebrigtsen <larsi@gnus.org> writes:
>
>     Lars> Dan Christensen <jdc@uwo.ca> writes:
>     >> It just occurred to me that some people use group parameters to
>     >> have different expiry-wait periods for different groups.  Will
>     >> this above cause articles to expire based on the expiry-wait for
>     >> the original group or whatever expiry-wait applies to the nnir
>     >> group (probably the default global value)?
>
>     Lars> The latter, I think.  But the nnir groups are ephemeral,
>     Lars> aren't they?  So they shouldn't affect anything.  Unless I'm
>     Lars> misunderstanding everything.
>
> My message crossed yours. This is correct: ephemeral, so it shouldn't
> cause any trouble (and hasn't in my one week of minimal testing).

I thought the whole point of this thread was that deletion in an nnir
group *does* cause deletion of the original article.  So I am assuming
that that holds as well for articles marked as expirable in an nnir
group.

So, say I do an nnir search, and the results contain an article in my
INBOX, which has a long expiry-wait.  I mark it as expirable in the nnir
group, think that that will just cause the mark to propagate to the
INBOX, where the usual expiry process will happen.  But when I exit the
nnir group, expiry runs, and the article disappears sooner than I
expected.

Dan

Re: deleting articles from search groups?

[Andrew Cohen]

>>>>> "Dan" == Dan Christensen <jdc@uwo.ca> writes:


[...]


    Dan> I thought the whole point of this thread was that deletion in
    Dan> an nnir group *does* cause deletion of the original article.
    Dan> So I am assuming that that holds as well for articles marked as
    Dan> expirable in an nnir group.

Yes. I think we are all saying correct things but talking about slightly
different issues. I brought the question up (really just about deleting
articles) precisely because the search group is ephemeral, and the user
might then be surprised that, unlike most ephemeral groups, certain
actions have permanent effects. This started with article moving, but
now includes article deleting as well.

Most users probably won't care about the ephemeral:non-ephemeral
distinction, and the ability to move and delete is so useful I'm not
worried about this anymore. But expiry (as opposed to forced deletion)
is another matter. Since nnir groups are ephemeral the marks don't
persist. So you can't try to propagate an expiry mark to the original
article. I guess this could be a confusion for some users. I also don't
see a use-case for marking anything expired in an nnir buffer: since the
mark is not persistent, its effect is to just delete the article (or
not) at exit from the summary buffer. I don't think there is much value
to leaving this decision this one time to the expiry mechanism. So I am
about to push a change that only allows deletion and not other kinds of
expiry. If anyone has a good argument for making expiry available in
this one-shot kind of way let me know.

Re: deleting articles from search groups?

[Andrew Cohen]

>>>>> "Dan" == Dan Christensen <jdc@uwo.ca> writes:

    Dan> Andrew Cohen <cohen@andy.bu.edu> writes:
    >>>>>>> "Lars" == Lars Magne Ingebrigtsen <larsi@gnus.org> writes:
    >> 
    >> >> But this means that gnus-summary-expire-articles is called on
    >> the >> nnir group upon summary-buffer exit.
    >> 
    Lars> But this is only called with the "E"-marked articles.  I guess
    Lars> if the user sets the "E" mark on nnir articles, they probably
    Lars> know that it's in danger of being deleted.  But I don't think
    Lars> anybody will do that, really...
    >> 
    >> OK, I've been living with this for a week and haven't had
    >> anything really bad happen. So I've pushed this. Let me know if
    >> anything goes wrong.

    Dan> It just occurred to me that some people use group parameters to
    Dan> have different expiry-wait periods for different groups.  Will
    Dan> this above cause articles to expire based on the expiry-wait
    Dan> for the original group or whatever expiry-wait applies to the
    Dan> nnir group (probably the default global value)?

The addition of nnir-request-expire-articles is that now the 'nnir group
articles are put through the expiry process. This is unconnected with
properties of the articles in their original group. And since articles
in 'nnir summary buffers are not going to be marked expirable, no expiry
should happen. Except for explicit deletion which just asks for
particular articles to be deleted, not run through the whole expiry
process.