bug#13306: 24.2; [WishList] rmail-next-error-move functionality for occurs mode

bug#13306: 24.2; [WishList] rmail-next-error-move functionality for occurs mode

[Mark Lillibridge]

    I noticed that an (unannounced) new function, rmail-next-error-move,
got added to Rmail recently.  It allows sane use of grep with Rmail
files (e.g., go to the correct message properly narrowed) so long as the
Rmail files are already open in Rmail mode.

    This is not terribly useful most of the time; what would be much
more useful is similar functionality for occurs mode run on a Rmail
buffer that has been widened.  That is, it would be really nice to be
able to widen the Rmail buffer (C-c C-w) then do an occurs search, which
would let you see the results in context for all the messages at once
and easily jump between the occurrences.

- Mark

bug#13306: clarification

[Mark Lillibridge]

    In case it wasn't clear from the last message, occurs does not work
sanely with Rmail buffers currently even in version 24.2.  It takes you
to the occurrence point but there is no way to narrow the buffer so you
can do things like forward the found message.

    Perhaps a way to narrow a widened Rmail buffer around point would be
a useful independent function.  I'm wondering if such a function called
from the right occurs mode hook (assuming it exists) would suffice to
provide the functionality I'm asking for here.

- Mark
PS, note that both grep and the proposed occurs search the raw mbox file
not the results of decoding the messages; if someone wanted to go for
extra awesomeness, they might try and make an special Rmail search
command that produces an occurs buffer but only searches the decoded
messages (e.g., ignore nontext parts, work on base 64/printable
representations correctly).  This would be substantially slower, but
would avoid a lot of false positives.

bug#13306: clarification

[Eli Zaretskii]

> From: Mark Lillibridge <mdl@alum.mit.edu>
> Date: Sat, 29 Dec 2012 15:02:36 -0800
> 
> 
>     In case it wasn't clear from the last message, occurs does not work
> sanely with Rmail buffers currently even in version 24.2.  It takes you
> to the occurrence point but there is no way to narrow the buffer so you
> can do things like forward the found message.
> 
>     Perhaps a way to narrow a widened Rmail buffer around point would be
> a useful independent function.  I'm wondering if such a function called
> from the right occurs mode hook (assuming it exists) would suffice to
> provide the functionality I'm asking for here.
> 
> - Mark
> PS, note that both grep and the proposed occurs search the raw mbox file
> not the results of decoding the messages; if someone wanted to go for
> extra awesomeness, they might try and make an special Rmail search
> command that produces an occurs buffer but only searches the decoded
> messages (e.g., ignore nontext parts, work on base 64/printable
> representations correctly).  This would be substantially slower, but
> would avoid a lot of false positives.

Mairix solves all this, since the results of the search are
_messages_, not just lines of text, and these results are presented to
Rmail that decodes and displays them as usual.  Try it.

bug#13306: clarification

[Mark Lillibridge]

Eli Zaretskii <eliz@gnu.org> writes:

>  > From: Mark Lillibridge <mdl@alum.mit.edu>
>  > Date: Sat, 29 Dec 2012 15:02:36 -0800
>  > 
>  > 
>  >     In case it wasn't clear from the last message, occurs does not work
>  > sanely with Rmail buffers currently even in version 24.2.  It takes you
>  > to the occurrence point but there is no way to narrow the buffer so you
>  > can do things like forward the found message.
>  > 
>  >     Perhaps a way to narrow a widened Rmail buffer around point would be
>  > a useful independent function.  I'm wondering if such a function called
>  > from the right occurs mode hook (assuming it exists) would suffice to
>  > provide the functionality I'm asking for here.
>  > 
>  > - Mark
>  > PS, note that both grep and the proposed occurs search the raw mbox file
>  > not the results of decoding the messages; if someone wanted to go for
>  > extra awesomeness, they might try and make an special Rmail search
>  > command that produces an occurs buffer but only searches the decoded
>  > messages (e.g., ignore nontext parts, work on base 64/printable
>  > representations correctly).  This would be substantially slower, but
>  > would avoid a lot of false positives.
>  
>  Mairix solves all this, since the results of the search are
>  _messages_, not just lines of text, and these results are presented to
>  Rmail that decodes and displays them as usual.  Try it.

    I already use Mairix; I want to use occurs on its results in order
to see the results in context.  Presumably someone else agrees since
they implemented this functionality for the result of grep.  Also, I
don't know a easy way to restrict Mairix's search to one Rmail file.

- Mark

bug#13306: 24.2; [WishList] rmail-next-error-move functionality for occurs mode

[Eli Zaretskii]

> From: Mark Lillibridge <mdl@alum.mit.edu>
> Date: Sat, 29 Dec 2012 14:43:03 -0800
> 
> 
>     I noticed that an (unannounced) new function, rmail-next-error-move,
> got added to Rmail recently.  It allows sane use of grep with Rmail
> files (e.g., go to the correct message properly narrowed) so long as the
> Rmail files are already open in Rmail mode.
> 
>     This is not terribly useful most of the time; what would be much
> more useful is similar functionality for occurs mode run on a Rmail
> buffer that has been widened.  That is, it would be really nice to be
> able to widen the Rmail buffer (C-c C-w) then do an occurs search, which
> would let you see the results in context for all the messages at once
> and easily jump between the occurrences.

Why don't you use mairix.el instead?  Then you could search _all_ of
your Rmail files, not just the inbox, and have the results presented
in a separate Rmail file.  The search is also more sophisticated and
flexible than just grep.  Take a look at the mairix manual for the
details.

IOW, instead of introducing one more feature, use an existing one.  It
serves me well, FWIW.