Marks added in nnir search buffers don't propagate to the actual groups?

Marks added in nnir search buffers don't propagate to the actual groups?

[Eric Abrahamsen]

Can anyone confirm that changes to marks made in an nnir search buffer
don't propagate to the real articles? Eg you search for an article,
remove its read mark, and then expect to find it unread in the "real"
underlying group. Or you tick it, or whatever.

This is something I noticed ages ago, but sort of figured it was a
problem in my config, but now I've looked into it and so far as I can
tell this is just the way it works.

When I search an nnimap group, and tick an article in the resulting nnir
summary buffer, I can see that nnir asks the nnimap backend about
updating the mark, but the mark is never actually set.
`gnus-summary-exit' is supposed to apply all marks to the backend, but
it very carefully doesn't do that for ephemeral groups (and that code is
ancient).

I sure thought this used to work, though. Maybe nnir used to apply marks
immediately (ie, not waiting for summary exit), but doesn't anymore? Am
I just dreaming? Does anyone have any idea of how this is supposed to
work?

Thanks,
Eric

Re: Marks added in nnir search buffers don't propagate to the actual groups?

[Gijs Hillenius]

On 28 November 2018 11:05 Eric Abrahamsen, wrote:

> Can anyone confirm that changes to marks made in an nnir search buffer
> don't propagate to the real articles? Eg you search for an article,
> remove its read mark, and then expect to find it unread in the "real"
> underlying group. Or you tick it, or whatever.
>
> This is something I noticed ages ago, but sort of figured it was a
> problem in my config, but now I've looked into it and so far as I can
> tell this is just the way it works.

This is also how I understand Gnus. I nearly automatically do A W
gnus-warp-to-article to set or remove tickmarks

Re: Marks added in nnir search buffers don't propagate to the actual groups?

[Eric S Fraga]

On Thursday, 29 Nov 2018 at 09:35, Gijs Hillenius wrote:
> On 28 November 2018 11:05 Eric Abrahamsen, wrote:
>
>> Can anyone confirm that changes to marks made in an nnir search buffer
>> don't propagate to the real articles? Eg you search for an article,
>> remove its read mark, and then expect to find it unread in the "real"
>> underlying group. Or you tick it, or whatever.
>>
>> This is something I noticed ages ago, but sort of figured it was a
>> problem in my config, but now I've looked into it and so far as I can
>> tell this is just the way it works.
>
> This is also how I understand Gnus. I nearly automatically do A W
> gnus-warp-to-article to set or remove tickmarks

Is it also the case generally for virtual groups?  Seems like it is but
it would good if it were not.  I would like marks to propagate to the
actual email wherever it might lie.
-- 
Eric S Fraga via Emacs 27.0.50 & org 9.1.13 on Debian 9.5

Re: Marks added in nnir search buffers don't propagate to the actual groups?

[Eric Abrahamsen]

Eric S Fraga <e.fraga@ucl.ac.uk> writes:

> On Thursday, 29 Nov 2018 at 09:35, Gijs Hillenius wrote:
>> On 28 November 2018 11:05 Eric Abrahamsen, wrote:
>>
>>> Can anyone confirm that changes to marks made in an nnir search buffer
>>> don't propagate to the real articles? Eg you search for an article,
>>> remove its read mark, and then expect to find it unread in the "real"
>>> underlying group. Or you tick it, or whatever.
>>>
>>> This is something I noticed ages ago, but sort of figured it was a
>>> problem in my config, but now I've looked into it and so far as I can
>>> tell this is just the way it works.
>>
>> This is also how I understand Gnus. I nearly automatically do A W
>> gnus-warp-to-article to set or remove tickmarks
>
> Is it also the case generally for virtual groups?  Seems like it is but
> it would good if it were not.  I would like marks to propagate to the
> actual email wherever it might lie.

Thanks for both your responses. It finally occurred to me, long after
the fact, to look at the "nnir" section of the manual, and it does
explicitly say that the groups are ephemeral and marks won't propagate
to the backend. Oh well. I sure thought they once did, though.

"virtual" groups aren't the same as "ephemeral" groups -- marks set in
virtual groups are propagated to the backends, but not those set in
ephemeral groups.

(Which begs the question of why there's a `nnir-request-set-mark'
function, but oh well.)

Thanks,
Eric

Re: Marks added in nnir search buffers don't propagate to the actual groups?

[Eric S Fraga]

On Thursday, 29 Nov 2018 at 09:41, Eric Abrahamsen wrote:
> "virtual" groups aren't the same as "ephemeral" groups -- marks set in
> virtual groups are propagated to the backends, but not those set in
> ephemeral groups.

Thanks for this.  I'll need to investigate further as I definitely
convinced myself that some marks, E in particular, did not seem to be
recorded in the actual article's group.
-- 
Eric S Fraga via Emacs 27.0.50 & org 9.1.13 on Debian 9.5