Virtual folders in macros

[Suvayu Ali]

Hi,

I would like to write a few macros for easy access to virtual folders.
Of course I can always create a query and put it in a macro, but then I
do not have the nice relevant names.  To give you an example:

virtual-mailboxes  "Topic"      "<complicated query>"

When visiting the above folder, I have the easy to read name "Topic" in
the status line.  But for the following macro:

macro index,pager gT "<vfolder-from-query><complicated query><enter>" \
       		       "Topic description for help screen"

I have "$nm_default_uri?query=<<complicated query>" as the folder name.
This can be a bit confusing; I have to look at the query carefully to
find out what I'm looking at.

I also never figured out how to enter a text string to <change-vfolder>.
If that is possible, it would solve the above issue too.

Thanks for any ideas,

[Vladimir Marek]

Hi,

I'm not saying I can do anything about it, I just want to clear up your
question.
> I would like to write a few macros for easy access to virtual folders.

First question, why? I guess the standard way is to create
virtual-mailboxes and select the 'folder' I want to see. What advantage
has macro for you?
> Of course I can always create a query and put it in a macro, but then I
> do not have the nice relevant names.  To give you an example:
>
> virtual-mailboxes  "Topic"      "<complicated query>"
>
> When visiting the above folder, I have the easy to read name "Topic" in
> the status line.  But for the following macro:
>
> macro index,pager gT "<vfolder-from-query><complicated query><enter>" \
>        		       "Topic description for help screen"
>
> I have "$nm_default_uri?query=<<complicated query>" as the folder name.
> This can be a bit confusing; I have to look at the query carefully to
> find out what I'm looking at.

Understood, the query string could have another parameter 'name'
similarly to 'limit' and 'type'. Is it worth? Karel Zak or anyone
creating a patch would know :)
> I also never figured out how to enter a text string to <change-vfolder>.
> If that is possible, it would solve the above issue too.

Yes, that didn't work for me also, but it's quite some time I tried
that.

The problem I have with virtual-mailboxes is, that I can't search them
by their name. If I want to go to folder 'notmuch' for example, I would
like to
 - press X
 - press CR (that should not be necessary, I guess. But it is at the moment)
 - / ( for searching a name)
 - type 'notmuch' and <CR>

Unfortunately the searching looks into the mailbox specification, and
since the mailbox specification always starts with 'notmuch://?query...'
just jumps to next line.

Cheers

[Suvayu Ali]

Hi Vlad,

On Tue, Dec 03, 2013 at 10:02:50AM +0100, Vladimir Marek wrote:
> Hi,
>
> I'm not saying I can do anything about it, I just want to clear up your
> question.

Actually, I have been meaning to get into the mutt source for a while.
If you can point me to the relevant functions, I could try it myself.
> > I would like to write a few macros for easy access to virtual folders.
>
> First question, why? I guess the standard way is to create
> virtual-mailboxes and select the 'folder' I want to see. What advantage
> has macro for you?

Well, macros are faster!  Without a macro, I usually use the
sidebar-next/prev functions followed by sidebar-open.  Compared to that,
using a macro is just two keystrokes.  I use them for (virtual) folders
I visit very often.
> > Of course I can always create a query and put it in a macro, but then I
> > do not have the nice relevant names.  To give you an example:
> >
> > virtual-mailboxes  "Topic"      "<complicated query>"
> >
> > When visiting the above folder, I have the easy to read name "Topic" in
> > the status line.  But for the following macro:
> >
> > macro index,pager gT "<vfolder-from-query><complicated query><enter>" \
> >        		       "Topic description for help screen"
> >
> > I have "$nm_default_uri?query=<<complicated query>" as the folder name.
> > This can be a bit confusing; I have to look at the query carefully to
> > find out what I'm looking at.
>
> Understood, the query string could have another parameter 'name'
> similarly to 'limit' and 'type'. Is it worth? Karel Zak or anyone
> creating a patch would know :)

I'm not sure; I thought the query string is supposed to be the Xapian
query passed on to libnotmuch.  This is more a UI enhancement, I would
think it is better to keep the two separate.
> > I also never figured out how to enter a text string to <change-vfolder>.
> > If that is possible, it would solve the above issue too.
>
> Yes, that didn't work for me also, but it's quite some time I tried
> that.
>
> The problem I have with virtual-mailboxes is, that I can't search them
> by their name. If I want to go to folder 'notmuch' for example, I would
> like to
> press X
> press CR (that should not be necessary, I guess. But it is at the moment)
> / ( for searching a name)
> type 'notmuch' and <CR>
>
> Unfortunately the searching looks into the mailbox specification, and
> since the mailbox specification always starts with 'notmuch://?query...'
> just jumps to next line.

This I do not follow.  This is what I see after 'X <any character>':

186(   137) Today
    236(   178) Yest
   3233(     0) Inboxes
    447(   447) virtual folder 1
   1330(   747) virtual folder 2

Now I can search with '/Today<enter>' and I'm in today.  How do you
define the virtual folders?  This is what I do:

virtual-mailboxes \
    "Inboxes"      "$my_nmuri is:inbox" \
    "Today"        "$my_nmuri date:today..today &limit=500" \
    "Yest"         "$my_nmuri date:yesterday..yesterday &limit=500"

The first quoted string serves as the folder name shown above.

Cheers,

[Vladimir Marek]

[ Rest was IMO answered by Karel already ]
> > The problem I have with virtual-mailboxes is, that I can't search them
> > by their name. If I want to go to folder 'notmuch' for example, I would
> > like to
> > press X
> > press CR (that should not be necessary, I guess. But it is at the moment)
> > / ( for searching a name)
> > type 'notmuch' and <CR>
> >
> > Unfortunately the searching looks into the mailbox specification, and
> > since the mailbox specification always starts with 'notmuch://?query...'
> > just jumps to next line.
>
> This I do not follow.  This is what I see after 'X <any character>':
> 186(   137) Today
> 236(   178) Yest
>
> 3233(     0) Inboxes
>     447(   447) virtual folder 1
>    1330(   747) virtual folder 2
>
> Now I can search with '/Today<enter>' and I'm in today.  How do you
> define the virtual folders?  This is what I do:
>
> virtual-mailboxes \
>     "Inboxes"      "$my_nmuri is:inbox" \
>     "Today"        "$my_nmuri date:today..today &limit=500" \
>     "Yest"         "$my_nmuri date:yesterday..yesterday &limit=500"
>
> The first quoted string serves as the folder name shown above.

It will work if you search for 'Today', but try searching for 'notmuch'
(I guess your $my_nmuri is something like 'notmuch://'). It does not
search the virtual mailbox name, but rather it's definition.

Tis is part of my mailbox definitions:

virtual-mailboxes \
  INBOX               "notmuch://?query=(tag:inbox and (tag:unread or tag:flagged) and not tag:unsorted)&type=threads"        \
  vm-interest         "notmuch://?query=tag:vm-interest and NOT tag:archive"        \
  vm-admin            "notmuch://?query=tag:vm-admin and NOT tag:archive"        \

[Suvayu Ali]

On Wed, Dec 04, 2013 at 04:12:59PM +0100, Vladimir Marek wrote:
> [ Rest was IMO answered by Karel already ]
> > > The problem I have with virtual-mailboxes is, that I can't search them
> > > by their name. If I want to go to folder 'notmuch' for example, I would
> > > like to
> > > press X
> > > press CR (that should not be necessary, I guess. But it is at the moment)
> > > / ( for searching a name)
> > > type 'notmuch' and <CR>
> > >
> > > Unfortunately the searching looks into the mailbox specification, and
> > > since the mailbox specification always starts with 'notmuch://?query...'
> > > just jumps to next line.
> >
> > This I do not follow.  This is what I see after 'X <any character>':
> > 186(   137) Today
> > 236(   178) Yest
> >
> > 3233(     0) Inboxes
> >     447(   447) virtual folder 1
> >    1330(   747) virtual folder 2
> >
> > Now I can search with '/Today<enter>' and I'm in today.  How do you
> > define the virtual folders?  This is what I do:
> >
> > virtual-mailboxes \
> >     "Inboxes"      "$my_nmuri is:inbox" \
> >     "Today"        "$my_nmuri date:today..today &limit=500" \
> >     "Yest"         "$my_nmuri date:yesterday..yesterday &limit=500"
> >
> > The first quoted string serves as the folder name shown above.
>
> It will work if you search for 'Today', but try searching for 'notmuch'
> (I guess your $my_nmuri is something like 'notmuch://'). It does not

Indeed, I use my_* variables to make my config more readable.
> search the virtual mailbox name, but rather it's definition.
>
> Tis is part of my mailbox definitions:
>
> virtual-mailboxes \
>   INBOX               "notmuch://?query=(tag:inbox and (tag:unread or tag:flagged) and not tag:unsorted)&type=threads"        \
>   vm-interest         "notmuch://?query=tag:vm-interest and NOT tag:archive"        \
>   vm-admin            "notmuch://?query=tag:vm-admin and NOT tag:archive"        \

Indeed, you are right!  My inbox example worked because the notmuch URI
also had inbox in it.  I tried a few others with no commonality between
the URI and the vfolder description, the searches failed.  One example:

"me @ Analysis" "$my_nmuri from:suvayu and folder:gwt"

Searching for ana fails to find the above.

Cheers,

[Karel Zak]

On Tue, Dec 03, 2013 at 10:02:50AM +0100, Vladimir Marek wrote:
> The problem I have with virtual-mailboxes is, that I can't search them
> by their name. If I want to go to folder 'notmuch' for example, I would
> like to
> press X
> press CR (that should not be necessary, I guess. But it is at the moment)
> / ( for searching a name)
> type 'notmuch' and <CR>
>
> Unfortunately the searching looks into the mailbox specification, and
> since the mailbox specification always starts with 'notmuch://?query...'
> just jumps to next line.

Fixed, not it uses folder name rather than the path.

Karel

[Vladimir Marek]

> > The problem I have with virtual-mailboxes is, that I can't search them
> > by their name. If I want to go to folder 'notmuch' for example, I would
> > like to
> > press X
> > press CR (that should not be necessary, I guess. But it is at the moment)
> > / ( for searching a name)
> > type 'notmuch' and <CR>
> >
> > Unfortunately the searching looks into the mailbox specification, and
> > since the mailbox specification always starts with 'notmuch://?query...'
> > just jumps to next line.
>
> Fixed, not it uses folder name rather than the path.

Nice! Thank you. Also now I can do

macro index X "<change-vfolder>?"

And pressing X always just presents list of vfolders.

Thank you second time.

[Vladimir Marek]

> > Unfortunately the searching looks into the mailbox specification, and
> > since the mailbox specification always starts with 'notmuch://?query...'
> > just jumps to next line.
>
> Fixed, not it uses folder name rather than the path.

Actually I remembered that there is another issue with vfolders -
folder-hook. Folder hook operates on 'path' of a folder, not on it's
description. That does not matter too much for other types of mailboxes
as they are based on real path somewhere. But notmuch's 'path' can be
very different. It would be great to have that changed too.

Looking at the sources, it won't be so easy. There are several ways of
entering a (v)folder.

- main.c:1022 mutt_select_file (folder, sizeof (folder), M_SEL_FOLDER | M_SEL_BUFFY);
 ( that's <change-vfolder>? I believe )

- main.c:647 case 'f':
           strfcpy (folder, optarg, sizeof (folder));
 ( that's $ mutt -f notmuch://... - I have never tried that!)

- calling <change-vfolder> directly
  ( somewhere in curs_main.c ? )

If we somehow cover all those entry points, setting not only 'folder'
variable but also new 'description' variable, we the could modify the

main.c:1051    mutt_folder_hook (folder);

to something like

if (we are dealing with vfolder)  {
    mutt_folder_hook (description);
} else {
mutt_folder_hook (folder);
}

How does that sound?

Cheers

[Karel Zak]

On Thu, Dec 05, 2013 at 11:30:36PM +0100, Vladimir Marek wrote:
> Actually I remembered that there is another issue with vfolders -
> folder-hook. Folder hook operates on 'path' of a folder, not on it's
> description. That does not matter too much for other types of mailboxes
> as they are based on real path somewhere. But notmuch's 'path' can be
> very different. It would be great to have that changed too.

mutt uses the $folder also for imap and pop where the path is very 
tricky (e.g. imap[s]:[user[:pw]@]imapserver.example.com[:port]/path).

IMHO it would be better to keep it the same for notmuch. The vfolder
description is just UI stuff to make things more user-friendly.

BTW, keep folder-hooks based on paths means that you can use hooks
also for on the fly generated vfolders (<vfolder-from-query>).
> Looking at the sources, it won't be so easy. There are several ways of
> entering a (v)folder.
> main.c:1022 mutt_select_file (folder, sizeof (folder), M_SEL_FOLDER | M_SEL_BUFFY);
>
> ( that's <change-vfolder>? I believe )

Yep
> main.c:647 case 'f':
>        strfcpy (folder, optarg, sizeof (folder));
>
> ( that's $ mutt -f notmuch://... - I have never tried that!)

it works, try:

mutt -f "notmuch://?query=from:@oracle.com"

the nice thing is that mutt has functions to detect folder (context)
type from path string, so add notmuch was relatively simple as it is
extension to the existing code only. Nowhere in this code we use the
description field.
> calling <change-vfolder> directly
>
> ( somewhere in curs_main.c ? )
>
> If we somehow cover all those entry points, setting not only 'folder'
> variable but also new 'description' variable, we the could modify the

Yes, it would be probably better to introduce a new variable than
play nasty games with $folder, then we can support things like

folder-hook notmuch:.*tag=FOO.* 'set folder_description="Something about FOO"'

Frankly, now the description is only stored in VirtIncoming (which is
parsed "virtual-mailboxes"). The description is not regular part of the
mutt Context (struct with the current folder). Maybe it would be
better to add the description to the Context and make it usable for
arbitrary folders (maildirs, mbox, ..) by "set folder_description".
> if (we are dealing with vfolder)  {
>    mutt_folder_hook (description);
> } else {
> mutt_folder_hook (folder);

I'd like to avoid this.

Karel

[Vladimir Marek]

> > Actually I remembered that there is another issue with vfolders -
> > folder-hook. Folder hook operates on 'path' of a folder, not on it's
> > description. That does not matter too much for other types of mailboxes
> > as they are based on real path somewhere. But notmuch's 'path' can be
> > very different. It would be great to have that changed too.
>
> mutt uses the $folder also for imap and pop where the path is very 
> tricky (e.g. imap[s]:[user[:pw]@]imapserver.example.com[:port]/path).

That is ture, but there's at least clear '/path' component which we
don't have.
> IMHO it would be better to keep it the same for notmuch. The vfolder
> description is just UI stuff to make things more user-friendly.

I don't have strong feelings about this, it was just something that was
slightly bothering me in the past, but nothing horrible.
> BTW, keep folder-hooks based on paths means that you can use hooks
> also for on the fly generated vfolders (<vfolder-from-query>).

Good point.
> > calling <change-vfolder> directly
> >
> > ( somewhere in curs_main.c ? )
> >
> > If we somehow cover all those entry points, setting not only 'folder'
> > variable but also new 'description' variable, we the could modify the
>
> Yes, it would be probably better to introduce a new variable than
> play nasty games with $folder, then we can support things like
>
> folder-hook notmuch:.*tag=FOO.* 'set folder_description="Something about FOO"'
>
> Frankly, now the description is only stored in VirtIncoming (which is
> parsed "virtual-mailboxes"). The description is not regular part of the
> mutt Context (struct with the current folder). Maybe it would be
> better to add the description to the Context and make it usable for
> arbitrary folders (maildirs, mbox, ..) by "set folder_description".

Maybe. On the other hand it would make potential merging back to mutt
sources harder.
> > if (we are dealing with vfolder)  {
> >    mutt_folder_hook (description);
> > } else {
> > mutt_folder_hook (folder);
>
> I'd like to avoid this.

Fair enough.

For example I would like to jump automatically to last-entry when
entering my INBOX. At the moment I have

virtual-mailboxes \
    INBOX "notmuch://?query=(tag:inbox and (tag:unread or tag:flagged) and not tag:unsorted)&type=threads" \

folder-hook 'tag:inbox( |  |$)' 'exec last-entry'

The one of the spaces is really a tabulator. (That could be replaced by
\y, now that I look to mutt documentation.)

Also, as a workaround, I could change the inbox specification to include

... and not tag:this_is_my_inbox

and match the 'this_is_my_inbox' in folder hook. It is very unlikely
that I would really use such tag. That's very simple workaround. So
don't worry about it :)

Cheers

[Vladimir Marek]

> folder-hook 'tag:inbox( |  |$)' 'exec last-entry'
>
> The one of the spaces is really a tabulator. (That could be replaced by
> \y, now that I look to mutt documentation.)

Actually \y does not seem to be implemented in mutt sources ... > does
the trick.

[Suvayu Ali]

On Thu, Dec 05, 2013 at 11:06:53AM +0100, Karel Zak wrote:
> On Tue, Dec 03, 2013 at 10:02:50AM +0100, Vladimir Marek wrote:
> > The problem I have with virtual-mailboxes is, that I can't search them
> > by their name. If I want to go to folder 'notmuch' for example, I would
> > like to
> > press X
> > press CR (that should not be necessary, I guess. But it is at the moment)
> > / ( for searching a name)
> > type 'notmuch' and <CR>
> >
> > Unfortunately the searching looks into the mailbox specification, and
> > since the mailbox specification always starts with 'notmuch://?query...'
> > just jumps to next line.
>
> Fixed, not it uses folder name rather than the path.

Updated packages with both these changes available on copr now:

http://copr-fe.cloud.fedoraproject.org/coprs/fatka/mutt-kz/

:)

[Karel Zak]

On Fri, Dec 06, 2013 at 01:02:37AM +0100, Suvayu Ali wrote:
> Updated packages with both these changes available on copr now:
>
> http://copr-fe.cloud.fedoraproject.org/coprs/fatka/mutt-kz/

Cool, it would be nice to update the packages to v1.5.22.1-rc1 and
 maybe also add "Conflicts: mutt" to avoid collisions.

Karel

[Suvayu Ali]

On Fri, Dec 06, 2013 at 03:57:37PM +0100, Karel Zak wrote:
> On Fri, Dec 06, 2013 at 01:02:37AM +0100, Suvayu Ali wrote:
> > Updated packages with both these changes available on copr now:
> >
> > http://copr-fe.cloud.fedoraproject.org/coprs/fatka/mutt-kz/
>
> Cool, it would be nice to update the packages to v1.5.22.1-rc1 and
>  maybe also add "Conflicts: mutt" to avoid collisions.

Done, but 2 things:

1. Copr is behaving weirdly, it intermittently finishes the build but
   fails to create the repo (that is what happened for this build).  So
   yum install will get you the older rpm.  To get the latest, do:

# yum install http://copr-be.cloud.fedoraproject.org/results/fatka/mutt-kz/fedora-19-x86_6...

2. Do you think using Obsoletes instead of Conflicts would be better?
   With Obsoletes, if someone installs mutt-kz on a system with mutt
   installed, it should smoothly remove mutt and install mutt-kz instead
   of giving an error and failing (as it does now).

Cheers,

[Karel Zak]

On Fri, Dec 06, 2013 at 09:02:08PM +0100, Suvayu Ali wrote:
> Do you think using Obsoletes instead of Conflicts would be better?
> With Obsoletes, if someone installs mutt-kz on a system with mutt
> installed, it should smoothly remove mutt and install mutt-kz instead
> of giving an error and failing (as it does now).

Hmm... OK, good point, let's use Obsoletes :-)

Karel

[Suvayu Ali]

On Tue, Dec 10, 2013 at 10:40:46AM +0100, Karel Zak wrote:
> On Fri, Dec 06, 2013 at 09:02:08PM +0100, Suvayu Ali wrote:
> > Do you think using Obsoletes instead of Conflicts would be better?
> > With Obsoletes, if someone installs mutt-kz on a system with mutt
> > installed, it should smoothly remove mutt and install mutt-kz instead
> > of giving an error and failing (as it does now).
>
> Hmm... OK, good point, let's use Obsoletes :-)

Done!  Repo creation failed again though.  People can find the rpm here:

http://copr-be.cloud.fedoraproject.org/results/fatka/mutt-kz/fedora-19-x86_6...

Cheers,

[Karel Zak]

On Tue, Dec 10, 2013 at 01:06:26PM +0100, Suvayu Ali wrote:
> Done!  Repo creation failed again though.  People can find the rpm here:
>
> [...] mutt-kz-1.5.22-1.20131210.git.d9309dab

Please, be more careful with versioing
https://fedoraproject.org/wiki/Packaging:NamingGuidelines#Package_Versioning

<name>-<version>-<release>

* <version> should be based on upstream version, the current version
  is "1.5.22.1"

* <release> is .spec file specific, if you create rpms for pre-release
  (beta, -rc, ..) packages then use ".0" prefix in the <release>

For example:

mutt-kz-1.5.22.1-0.20131210gitd9309dab

(and maybe the hash "d9309dab" is unnecessary as you probably don't create more
  than one snapshot per day:-)

in .spec file:

Version: 1.5.22.1
   Release: 0.20131210gitd9309dab

Anyway, we have tarballs at ftp://redcrew.org/pub/mutt-kz/v1.5.22.1/ so you don't
have to use snapshots from git.

All you need is:

Version: 1.5.22.1
   Release: 0.1

%define upstream_version %{version}-rc1

Source: ftp://redcrew.org/pub/mutt-kz/v%{version}/mutt-kz-%{upstream_version}.tar.gz

%prep
   %setup -q -n %{name}-%{upstream_version}

to make it easy.

Thanks!

Karel

[Suvayu Ali]

Hi Karel,

Firstly, apologies for the delay; I got really busy with other things,
and then the holidays.

On Mon, Dec 16, 2013 at 11:03:22AM +0100, Karel Zak wrote:
> On Tue, Dec 10, 2013 at 01:06:26PM +0100, Suvayu Ali wrote:
> > Done!  Repo creation failed again though.  People can find the rpm here:
> >
> > [...] mutt-kz-1.5.22-1.20131210.git.d9309dab
>
> Please, be more careful with versioing
> https://fedoraproject.org/wiki/Packaging:NamingGuidelines#Package_Versioning
>
> <name>-<version>-<release>

I think I have fixed all the issues.  This is the package version number
after the changes: 5:1.5.22.1-0.1.rc1.20131222.git.2f3e0f76.  You can
find the latest package on copr.  If you had installed the previous
package (with faulty release tag) by hand, please uninstall it before
using yum update; things should work now nicely.
> mutt-kz-1.5.22.1-0.20131210gitd9309dab
>
> (and maybe the hash "d9309dab" is unnecessary as you probably don't create more
>   than one snapshot per day:-)

Using release tarballs is very convenient for packaging, but I would
prefer to keep using git snapshots.  I want the packages to include the
latest fixes you and others put in, they get more testing this way :).

I hope that is okay.

Cheers,

[Karel Zak]

On Mon, Dec 02, 2013 at 06:03:36PM +0100, Suvayu Ali wrote:
> I would like to write a few macros for easy access to virtual folders.
> Of course I can always create a query and put it in a macro, but then I
> do not have the nice relevant names.  To give you an example:
>
> virtual-mailboxes  "Topic"      "<complicated query>"
>
> When visiting the above folder, I have the easy to read name "Topic" in
> the status line.  But for the following macro:
>
> macro index,pager gT "<vfolder-from-query><complicated query><enter>" \
>        		       "Topic description for help screen"
>
> I have "$nm_default_uri?query=<<complicated query>" as the folder name.

The mutt searches for the description text (e.g. "Topic") by the query
 URI, so if the folder URI is the same as URI used in the macro then
 you will see the description text in the status line.

Anyway, I don't think that define a new vfolder by macro is the right
 way. It would be better to keep the definitions in "virtual-mailboxes" 
 and have a way how to address the definitions.

I'll try to play with that, it seems that the current <change-vfolder> is
 limited (compare to regular change-folder).
> I also never figured out how to enter a text string to <change-vfolder>.
> If that is possible, it would solve the above issue too.

Yep.

Karel

[Karel Zak]

On Mon, Dec 02, 2013 at 06:03:36PM +0100, Suvayu Ali wrote:
> I also never figured out how to enter a text string to <change-vfolder>.
> If that is possible, it would solve the above issue too.

Fixed, git pull. Now you can specify the folder by name or URI.

Karel

[Vladimir Marek]

> > I also never figured out how to enter a text string to <change-vfolder>.
> > If that is possible, it would solve the above issue too.
>
> Fixed, git pull. Now you can specify the folder by name or URI.

What took you so long? Whole 3 minutes!

Thank you :)

[SGT. Garcia]

From: glphvgacs 
To: [e-mail]
Cc: 
Bcc: 
Subject: vsubfolders [was: Virtual folders in macros]
Reply-To: 
In-Reply-To:

On Wed, Dec 04, 2013 at 01:02:00PM +0100, Karel Zak wrote:
> On Mon, Dec 02, 2013 at 06:03:36PM +0100, Suvayu Ali wrote:
> > I also never figured out how to enter a text string to <change-vfolder>.
> > If that is possible, it would solve the above issue too.
>
> Fixed, git pull. Now you can specify the folder by name or URI.
> Karel

just built it and noticed something.

btw: this might already be possible to do, i mind you, but i just don't
know how. so pointers in that regard are welcome.

so i have these vfolders defined:

"5665     [   -->   ]" "notmuch://?query=folder:5665 AND tag:unread NOT from:[e-mail]" \
  "         [  <--    ]" "notmuch://?query=from:[e-mail]" \
  "         [  <-->   ]" "notmuch://?query=folder:5665 AND from:[e-mail] AND to:[e-mail]" \
  "         [   ..    ]" "notmuch://?query=folder:5665" \
  "drw      [   -->   ]" "notmuch://?query=folder:drw AND tag:unread NOT from:[e-mail]&limit=32" \
  "         [  <--    ]" "notmuch://?query=from:[e-mail]" \
  "         [  <-->   ]" "notmuch://?query=folder:drw AND from:[e-mail] AND to:[e-mail]" \
  "         [   ..    ]" "notmuch://?query=folder:drw" \
and a macro to emulate gmail's key binding:
macro index gs "<change-vfolder>         [  <--    ]<enter>" "Go to Sent"

now if i'm at "drw      [   -->   ]" and hit 'gs' this will take me to:
  "         [  <--    ]" "notmuch://?query=from:[e-mail]" \
whereas one would expect to be landed at:
  "         [  <--    ]" "notmuch://?query=from:[e-mail]" \

i suppose the way it does it now is "query from the top" as opposed to
"change your frame of reference with respect to where you are then govern
yourself respectively".
btw, even if you change to query-from-here-on i don't think it would
completely solve the problem since i, and i'm sure there users out there
doing it, have another macro to land me in *INBOX*:
macro index gi "<change-vfolder>drw      [   -->   ]<enter>" "Go to INBOX"

right now it's not even possible to change-vfolder *dynamically* since
'drw' is hard-coded. and if it does query-from-here-on there are case where
it would get it wrong. say user does a 'gi' at:
  "         [  <--    ]" "notmuch://?query=from:[e-mail]" \

that will change to drw's INBOX and not 5665's.

so a suggestion would be to add the notion of vsubfolders that are somehow associated to a master vfolder.
i can elaborate if that interests you.

[Karel Zak]

On Wed, Dec 04, 2013 at 11:23:25AM -0500, glphvgacs wrote:
> so i have these vfolders defined:
>
> "5665     [   -->   ]" "notmuch://?query=folder:5665 AND tag:unread NOT from:[e-mail]" \
>   "         [  <--    ]" "notmuch://?query=from:[e-mail]" \
>   "         [  <-->   ]" "notmuch://?query=folder:5665 AND from:[e-mail] AND to:[e-mail]" \
>   "         [   ..    ]" "notmuch://?query=folder:5665" \
>   "drw      [   -->   ]" "notmuch://?query=folder:drw AND tag:unread NOT from:[e-mail]&limit=32" \
>   "         [  <--    ]" "notmuch://?query=from:[e-mail]" \
>   "         [  <-->   ]" "notmuch://?query=folder:drw AND from:[e-mail] AND to:[e-mail]" \
>   "         [   ..    ]" "notmuch://?query=folder:drw" \
> and a macro to emulate gmail's key binding:
> macro index gs "<change-vfolder>         [  <--    ]<enter>" "Go to Sent"
>
> now if i'm at "drw      [   -->   ]" and hit 'gs' this will take me to:
>   "         [  <--    ]" "notmuch://?query=from:[e-mail]" \
> whereas one would expect to be landed at:
>   "         [  <--    ]" "notmuch://?query=from:[e-mail]" \
>
> i suppose the way it does it now is "query from the top" as opposed to
> "change your frame of reference with respect to where you are then govern
> yourself respectively".
> btw, even if you change to query-from-here-on i don't think it would

It should not be so big problem to implement query-from-here-on, but
I'm not sure it's really so important as people almost always use
unique names for the folders. Frankly, your setup is very unusual :-)
> completely solve the problem since i, and i'm sure there users out there
> doing it, have another macro to land me in *INBOX*:
> macro index gi "<change-vfolder>drw      [   -->   ]<enter>" "Go to INBOX"
>
> right now it's not even possible to change-vfolder *dynamically* since
> 'drw' is hard-coded. and if it does query-from-here-on there are case where
> it would get it wrong. say user does a 'gi' at:
>   "         [  <--    ]" "notmuch://?query=from:[e-mail]" \
>
> that will change to drw's INBOX and not 5665's.
>
> so a suggestion would be to add the notion of vsubfolders that are somehow associated to a master vfolder.
> i can elaborate if that interests you.

so <change-vfolder> will go to the next master folder and <change-vfolder-sub>
 will go to the next sub-vfolder within the current master, right?

The idea of sub-vfolder is not so crazy (for example I already use "RH" and 
 "RH-<something>" to make it more obvious in sidebar).

Karel

[SGT. Garcia]

On Thu, Dec 05, 2013 at 11:27:49AM +0100, Karel Zak wrote:
> It should not be so big problem to implement query-from-here-on, but I'm
> not sure it's really so important as people almost always use unique
> names for the folders. Frankly, your setup is very unusual :-)

maybe because i'm using mutt for multiple accounts (gmail, .edu and so on).
it would be interesting to know if people in the community use it that way.
> so <change-vfolder> will go to the next master folder and
>  <change-vfolder-sub> will go to the next sub-vfolder within the current
>  master, right?

exact, plus...
> The idea of sub-vfolder is not so crazy (for example I already use "RH"
>  and "RH-<something>" to make it more obvious in sidebar).
> Karel

plus ^that. that is, also having sub-vfolders under their master vfolders
in the sidebar would be great too. in fact, IMHO, this should tolerate
recursion so one can even have accounts from same server bundled together.

say i have 3 accounts. 2 from gmail and one from my university. sidebar
would show something like:

edu
  INBOX
  sent
  recent
  all
  ...
gmail
  drw
    INBOX
    sent
    recent
    all
    ...
  5665
    INBOX
    sent
    recent
    all
    ...

and of course calls to notmuch would preferably be aware of those
associations,

[Suvayu Ali]

On Wed, Dec 04, 2013 at 01:02:00PM +0100, Karel Zak wrote:
> On Mon, Dec 02, 2013 at 06:03:36PM +0100, Suvayu Ali wrote:
> > I also never figured out how to enter a text string to <change-vfolder>.
> > If that is possible, it would solve the above issue too.
>
> Fixed, git pull. Now you can specify the folder by name or URI.

I just built it, works wonderfully.

PS: Copr is acting weird, build succeeded but failed to create a repo.
    I also had login problems for an hour or so.