notmuch-poll and OfflineIMAP
Jacek Generowicz writes: > Hello, > > Would you have any advice on how to construct a notmuch-poll script > that would work well in concert with OfflineIMAP? In particlular, how > can you avoid having to re-issue the IMAP account passwords? > > The OfflineIMAP docs state that SIGUSR1 forces an immediate resync of > all accounts, so something based around > > kill -SIGUSR1 `cat ~/.offlineimap/pid` > > or > > pkill -SIGUSR1 -u `whoami` offlineimap > > (with an already-running OfflineIMAP process in autorefresh mode) > could work, but it's not clear to me how to discover when the resync > is done, and sleeping for an arbitrary time before calling 'notmuch > new' seems unsatisfactory. > > Any words of wisdom? Offlineimap supports running hooks after syncing. Check if SIGUSR1 does that too. Then you could run notmuch new in the post sync hook. Also, notmuch new supports pre-new and post-new hooks. You could start or SIGUSR1 offlineimap in the pre-new hook. See man notmuch-hooks for details. HTH, Jani.
Passwordless multi-account SMTP
Hello, I'm trying to use notmuch from Emacs. I have a number of SMTP accounts through which I would need to send mail. Suggested solutions I have found fall into 3 categories: + Gnus posting styles. (I'd rather stay clear of gnus, at present.) + Writing plaintext passwords in the MSMTP config files. (I'm rather relutctant to leave plaintext passwords lying around.) + Emacs' own SMTP library. (I don't understand how to get it to deal with many accounts in a non-gnus based way. It also seems to be undergoing an interface change in the transition from Emacs 23 to 24, and it was already confusing enough without that complication.) To further complicate matters, I'm looking for a solution which queues any messages which are sent while the machine is offline. (It seems that MSMTP and Emacs SMTP library do offer some help in those respects.) Hints gratefully received. Thank you.
notmuch-poll and OfflineIMAP
Hello, Would you have any advice on how to construct a notmuch-poll script that would work well in concert with OfflineIMAP? In particlular, how can you avoid having to re-issue the IMAP account passwords? The OfflineIMAP docs state that SIGUSR1 forces an immediate resync of all accounts, so something based around kill -SIGUSR1 `cat ~/.offlineimap/pid` or pkill -SIGUSR1 -u `whoami` offlineimap (with an already-running OfflineIMAP process in autorefresh mode) could work, but it's not clear to me how to discover when the resync is done, and sleeping for an arbitrary time before calling 'notmuch new' seems unsatisfactory. Any words of wisdom? Thanks.
Passwordless multi-account SMTP
Hi Jacek, I am using msmtp for this: I keep my passwords in an encrypted file and use a shellscript to decrypt and grep for passwords in combination with msmtp's `passwordeval` option. Have a look at https://github.com/pazz/configs/blob/master/.msmtprc https://github.com/pazz/scripts/blob/master/getpwd I have a similar setup for offlineimap (configs also on github). All this works quite nicely together with my ssh-agent (gnome-keyring-daemon) which unlocks my gnupg keys upon login. HTH, /p Quoting Jacek Generowicz (2012-04-03 16:29:07) >Hello, > >I'm trying to use notmuch from Emacs. I have a number of SMTP accounts >through which I would need to send mail. Suggested solutions I have >found fall into 3 categories: > >+ Gnus posting styles. (I'd rather stay clear of gnus, at present.) > >+ Writing plaintext passwords in the MSMTP config files. (I'm rather > relutctant to leave plaintext passwords lying around.) > >+ Emacs' own SMTP library. (I don't understand how to get it to deal > with many accounts in a non-gnus based way. It also seems to be > undergoing an interface change in the transition from Emacs 23 to > 24, and it was already confusing enough without that complication.) > >To further complicate matters, I'm looking for a solution which queues >any messages which are sent while the machine is offline. (It seems >that MSMTP and Emacs SMTP library do offer some help in those >respects.) > >Hints gratefully received. > >Thank you. >___ >notmuch mailing list >notmuch at notmuchmail.org >http://notmuchmail.org/mailman/listinfo/notmuch
Message deletion wisdom
On Tue, Apr 03 2012, Jacek Generowicz wrote: > Looking through the archives of this list, I see that the topic of > message deletion is, if not controversial, then at least non-trivial. > > + Requests for the provision of message deletion mechanisms have > appeared many times, and that various patches addressing the issue > have been submitted, and (IIUC) some of these are awaiting review > and might possibly make it into an official release in the future. > > + Having the 'deleted' tag cause messages to be deleted is, > apparently, fraught with peril of losing mail. Hi, Jacek. You are right on both of these points. It has indeed been a controversial topic. However, I would say that we have reached a stasis in terms of our response to this issue. I'll try to summarize here. I can say with a very high degree of certainly that notmuch will NEVER gain the capability to actually delete mail files from disk itself. It is too risky for notmuch to be involved in that, and it's too easy to do it outside of notmuch (e.g. "notmuch search --output=files tag:deleted | xargs rm"). This has never really been up for discussion. What *has* generated a lot of discussion is whether or not the emacs interface should by default support key bindings that add a "deleted" tag to messages, and how that tag should be handled. I think we have finally settled on notmuch NOT supporting such a key binding by default, since it's easy enough for users to add the binding themselves if they wish. What we have added (or more accurately "are adding") though (thanks to the work of Austin Clements and Mark Walters) is the ability to exclude messages with a certain tag from search results. This is like an effective deletion, since messages can be completely omitted from search results without having them actually be deleted from disk. This is obviously much safer, and a more generally useful functionality. This functionality is currently in beta, but will be full-featured in the next release. So in summary, if you would like to "delete" messages, you can: * add a key binding to emacs (or your favorite ui) to add a "deleted" tag to messages that you want to delete: (define-key notmuch-show-mode-map "d" (lambda () (interactive) (notmuch-show-tag-message "+deleted"))) * add exclusion for messages with the "deleted" tag: notmuch config set search.exclude_tags deleted * and if you really want them purged from disk, delete them manually with: notmuch search --output=files tag:deleted | xargs -l rm I'll add a page about this to the wiki, since it comes up enough. hth. jamie. -- next part -- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 835 bytes Desc: not available URL: <http://notmuchmail.org/pipermail/notmuch/attachments/20120403/058920b8/attachment.pgp>
[PATCH 6/8] cli: add support for batch tagging operations to "notmuch tag"
On Tue, Apr 03 2012, Jani Nikula wrote: > Yes, it's more flexible, but what are the real benefits of such > flexibility? What other commands than tag/restore would truly benefit > from this? I might add that nobody has asked for such flexibility. But I do believe it was your patch that introduced the idea of batch processing of other commands to begin with. I think you're just turning my original question to you back on me. There's no point in adding and initial flag to the beginning of the line to specify the command to execute if you have no intention of ever supporting other commands. Or vice versa, if you ultimately want to support multiple commands then the place to do it is in the top level interface, and not in one of the subcommands. Ultimately, I think this patch series suffers from being simultaneously too close to existing functionality (restore), yet tantalizingly close to a much more generally useful concept of batch command processing. I would rather see it go one way or the other rather than add something fairly redundant in the middle. jamie. -- next part -- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 835 bytes Desc: not available URL: <http://notmuchmail.org/pipermail/notmuch/attachments/20120403/6099856e/attachment.pgp>
notmuch-poll and OfflineIMAP
Hi Jacek, On Tue, Apr 3, 2012 at 09:27, Jacek Generowicz wrote: > Would you have any advice on how to construct a notmuch-poll script > that would work well in concert with OfflineIMAP? ?In particlular, how > can you avoid having to re-issue the IMAP account passwords? > > The OfflineIMAP docs state that SIGUSR1 forces an immediate resync of > all accounts, so something based around > > ? ?kill -SIGUSR1 `cat ~/.offlineimap/pid` > > or > > ? ?pkill -SIGUSR1 -u `whoami` offlineimap > > (with an already-running OfflineIMAP process in autorefresh mode) > could work, but it's not clear to me how to discover when the resync > is done, and sleeping for an arbitrary time before calling 'notmuch > new' seems unsatisfactory. > > Any words of wisdom? I run offlineimap in autorefresh mode (with IDLE too, actually) and use a script based on inotify to update notmuch whenever new mail comes in: https://gist.github.com/1952483 . It sounds like some variant of this would probably work for you. As an added bonus, I have my post-new script run emacsclient -n -e '(notmuch-hello-update t)', so if I flip to my mail desktop I can see right away if there's new mail.
Handling of symlinked maildirs?
On Mon, 2 Apr 2012 10:54:48 -0700, Jameson Graef Rollins wrote: > On Mon, Apr 02 2012, Andrei POPESCU wrote: > > Im sorting my mailing lists with generic maildrop rules like this one: > > > > if (/^List-Id:.*/) > > to Maildir/.debian.$MATCH1/ > > > > However, I'm subscribed to a *lot* of mailing lists and in order to keep > > my folder view sane I use symlinks to conflate some of them, e.g. > > > > .debian.devel-announce -> .debian.devel > > > > This works well since mutt simply ignores the symlink(s) so I don't even > > need to exclude them in the config, but it seems that notmuch does index > > each of the symlinks as a separate folder[1]. > > > > Does it make sense to have this configurable or even completely exclude > > the symlinks? > > Hey, Andrei. I would say that if you have a compelling reason that > notmuch should be able to ignore symlinked directories then it maybe > makes sense to have it configurable. However, I must say that I have > completely dropped the whole folder concept myself since I find it > totally redundant with notmuch's tagging and searching capabilities. Please don't exclude the indexing of symlinks in notmuch. I have a fairly restricted IT environment at work, where I have chosen to symlink all my maildirs to other disks so that they appear in my home directory but do not fill my home disk quota every day. Often my different maildirs are to make my non-notmuch mail viewing slightly tolerable. There was a configuration option (new.ignore) added to 0.12 that lets one specify names of things not to index. That may be enough to do what you want, although not as automatic as just pitching all symlinks while indexing. Thanks, -Mark
Message deletion wisdom
Hello, Looking through the archives of this list, I see that the topic of message deletion is, if not controversial, then at least non-trivial. + Requests for the provision of message deletion mechanisms have appeared many times, and that various patches addressing the issue have been submitted, and (IIUC) some of these are awaiting review and might possibly make it into an official release in the future. + Having the 'deleted' tag cause messages to be deleted is, apparently, fraught with peril of losing mail. I have two questions: 1) Could you give me a summary of the dangers of allowing notmuch to actually delete mail? 2) How do you deal with mail that you would like to disappear from your maildirs/IMAP servers? (An Emacs-based slant would be particularle welcome.) Thank you for your input.
Passwordless multi-account SMTP
On Tue, 3 Apr 2012 17:29:07 +0200, Jacek Generowicz wrote: > Hello, > > I'm trying to use notmuch from Emacs. I have a number of SMTP accounts > through which I would need to send mail. Suggested solutions I have > found fall into 3 categories: > > + Gnus posting styles. (I'd rather stay clear of gnus, at present.) > > + Writing plaintext passwords in the MSMTP config files. (I'm rather > relutctant to leave plaintext passwords lying around.) > > + Emacs' own SMTP library. (I don't understand how to get it to deal > with many accounts in a non-gnus based way. It also seems to be > undergoing an interface change in the transition from Emacs 23 to > 24, and it was already confusing enough without that complication.) > > To further complicate matters, I'm looking for a solution which queues > any messages which are sent while the machine is offline. (It seems > that MSMTP and Emacs SMTP library do offer some help in those > respects.) > > Hints gratefully received. I use a slightly modified msmtpq for queuing and I have some hooks that run profiles to choose SMTP server and other options based on sender/receiver/last search. Passwords are plain text but my fs is encrypted. If you want, you could encrypt the password file with gpg and use gpg-agent to grab the password as needed from the file. Good luck. -- next part -- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 489 bytes Desc: not available URL: <http://notmuchmail.org/pipermail/notmuch/attachments/20120403/cbebe91b/attachment.pgp>
[PATCH 6/8] cli: add support for batch tagging operations to "notmuch tag"
On Mon, 02 Apr 2012 14:43:16 -0700, Jameson Graef Rollins wrote: > So what if we instead modified the top level binary ("notmuch") to: > > * add an option to specify that commands are to be processed from stdin > (--batch) > > * when in batch mode the db is opened once at the beginning and locked > > * commands are processed from stdin in the exact same form they are > specified on the command line ("tag +foo -- from:bar", "search > tag:foo", etc.). > > * db is closed when EOF is reached. > > That seems like it would be a generally much cleaner interface, and much > more flexible. Yes, it's more flexible, but what are the real benefits of such flexibility? What other commands than tag/restore would truly benefit from this? I might add that nobody has asked for such flexibility. As to the cleanliness of the interface, all is well as long as no characters needing escaping are used. With them, things get hairy. On the command line, the shell provides the necessary escaping and quoting mechanisms, and parsing to subcommand argv. I think it would be confusing to require %NN style hex escaping on the command line where shell mechanisms can be used, but also it would be a lot of work to duplicate the shell escaping and quoting mechanisms to the stdin parsing. For simplicity, I'd use hex escaping for stdin, and leave the command line to the shell. Finally, I'm not sure how the config and database open/close could be cleanly duplicated in top level "notmuch" and the subcommands. Should there be logic in the subcommands not to open the config/database if they're already opened in top level? All of notmuch cli should also be reworked to pass the config and database on to the subcommands. Not all the commands need to open the database in read/write mode, and some commands don't need to open the database at all, so it would not make sense to move config/database open/close to top level completely. In summary, I don't see enough value in this to put all the work required into it. I think it's way more trouble than it's worth. BR, Jani.
[PATCH 6/8] cli: add support for batch tagging operations to "notmuch tag"
On Apr 3, 2012 12:20 AM, "Jameson Graef Rollins" wrote: > > On Mon, Apr 02 2012, Jani Nikula wrote: > > Batch tagging brings performance and atomicity by opening and closing the > > db only once. The hex encoding handles insane message ids and tags. > > Otherwise there should be no difference. > > I can maybe see performance in terms of opening and closing the db only > once, but atomicity? Not sure I see how that is improved. Err, I meant locking, sorry. Nobody else is able to alter the db while batch tagging is in progress. > > And shouldn't we be able to improve the handling of command line > arguments so that we get the same encoding handling on the command line > as we would from stdin? > > jamie. -- next part -- An HTML attachment was scrubbed... URL: <http://notmuchmail.org/pipermail/notmuch/attachments/20120403/1e0695bf/attachment.html>
[PATCH 6/8] cli: add support for batch tagging operations to "notmuch tag"
On Apr 2, 2012 11:42 PM, "Jameson Graef Rollins" wrote: > > On Mon, Apr 02 2012, David Bremner wrote: > > Jameson Graef Rollins writes: > >> On Sat, Mar 31 2012, Jani Nikula wrote: > >>> Add support for batch tagging operations through stdin to "notmuch > >>> tag". This can be enabled with the new --stdin command line option to > >>> "notmuch new". The input must consist of lines of the format: > >>> > >>> T +|- [...] [--] > >> > > > > I think that's my fault. I was imagining a possible future line-oriented > > notmuch server and having various actions/queries possible. It seems a > > bit blue sky at this point, but it does give extesibility fairly > > cheaply. > > But then why not just make the command explicit, and just have the first > field be "tag"? > > But then I wonder why do we even need any of this at all? Isn't it this > just exactly equivalent to: > > xargs -l notmuch > ?? Batch tagging brings performance and atomicity by opening and closing the db only once. The hex encoding handles insane message ids and tags. Otherwise there should be no difference. Jani. > > jamie. -- next part -- An HTML attachment was scrubbed... URL: <http://notmuchmail.org/pipermail/notmuch/attachments/20120403/fb4fcd4d/attachment.html>
Re: [PATCH 6/8] cli: add support for batch tagging operations to notmuch tag
On Mon, 02 Apr 2012 14:43:16 -0700, Jameson Graef Rollins jroll...@finestructure.net wrote: So what if we instead modified the top level binary (notmuch) to: * add an option to specify that commands are to be processed from stdin (--batch) * when in batch mode the db is opened once at the beginning and locked * commands are processed from stdin in the exact same form they are specified on the command line (tag +foo -- from:bar, search tag:foo, etc.). * db is closed when EOF is reached. That seems like it would be a generally much cleaner interface, and much more flexible. Yes, it's more flexible, but what are the real benefits of such flexibility? What other commands than tag/restore would truly benefit from this? I might add that nobody has asked for such flexibility. As to the cleanliness of the interface, all is well as long as no characters needing escaping are used. With them, things get hairy. On the command line, the shell provides the necessary escaping and quoting mechanisms, and parsing to subcommand argv. I think it would be confusing to require %NN style hex escaping on the command line where shell mechanisms can be used, but also it would be a lot of work to duplicate the shell escaping and quoting mechanisms to the stdin parsing. For simplicity, I'd use hex escaping for stdin, and leave the command line to the shell. Finally, I'm not sure how the config and database open/close could be cleanly duplicated in top level notmuch and the subcommands. Should there be logic in the subcommands not to open the config/database if they're already opened in top level? All of notmuch cli should also be reworked to pass the config and database on to the subcommands. Not all the commands need to open the database in read/write mode, and some commands don't need to open the database at all, so it would not make sense to move config/database open/close to top level completely. In summary, I don't see enough value in this to put all the work required into it. I think it's way more trouble than it's worth. BR, Jani. ___ notmuch mailing list notmuch@notmuchmail.org http://notmuchmail.org/mailman/listinfo/notmuch
Message deletion wisdom
Hello, Looking through the archives of this list, I see that the topic of message deletion is, if not controversial, then at least non-trivial. + Requests for the provision of message deletion mechanisms have appeared many times, and that various patches addressing the issue have been submitted, and (IIUC) some of these are awaiting review and might possibly make it into an official release in the future. + Having the 'deleted' tag cause messages to be deleted is, apparently, fraught with peril of losing mail. I have two questions: 1) Could you give me a summary of the dangers of allowing notmuch to actually delete mail? 2) How do you deal with mail that you would like to disappear from your maildirs/IMAP servers? (An Emacs-based slant would be particularle welcome.) Thank you for your input. ___ notmuch mailing list notmuch@notmuchmail.org http://notmuchmail.org/mailman/listinfo/notmuch
notmuch-poll and OfflineIMAP
Hello, Would you have any advice on how to construct a notmuch-poll script that would work well in concert with OfflineIMAP? In particlular, how can you avoid having to re-issue the IMAP account passwords? The OfflineIMAP docs state that SIGUSR1 forces an immediate resync of all accounts, so something based around kill -SIGUSR1 `cat ~/.offlineimap/pid` or pkill -SIGUSR1 -u `whoami` offlineimap (with an already-running OfflineIMAP process in autorefresh mode) could work, but it's not clear to me how to discover when the resync is done, and sleeping for an arbitrary time before calling 'notmuch new' seems unsatisfactory. Any words of wisdom? Thanks. ___ notmuch mailing list notmuch@notmuchmail.org http://notmuchmail.org/mailman/listinfo/notmuch
Passwordless multi-account SMTP
Hello, I'm trying to use notmuch from Emacs. I have a number of SMTP accounts through which I would need to send mail. Suggested solutions I have found fall into 3 categories: + Gnus posting styles. (I'd rather stay clear of gnus, at present.) + Writing plaintext passwords in the MSMTP config files. (I'm rather relutctant to leave plaintext passwords lying around.) + Emacs' own SMTP library. (I don't understand how to get it to deal with many accounts in a non-gnus based way. It also seems to be undergoing an interface change in the transition from Emacs 23 to 24, and it was already confusing enough without that complication.) To further complicate matters, I'm looking for a solution which queues any messages which are sent while the machine is offline. (It seems that MSMTP and Emacs SMTP library do offer some help in those respects.) Hints gratefully received. Thank you. ___ notmuch mailing list notmuch@notmuchmail.org http://notmuchmail.org/mailman/listinfo/notmuch
Re: Passwordless multi-account SMTP
Hi Jacek, I am using msmtp for this: I keep my passwords in an encrypted file and use a shellscript to decrypt and grep for passwords in combination with msmtp's `passwordeval` option. Have a look at https://github.com/pazz/configs/blob/master/.msmtprc https://github.com/pazz/scripts/blob/master/getpwd I have a similar setup for offlineimap (configs also on github). All this works quite nicely together with my ssh-agent (gnome-keyring-daemon) which unlocks my gnupg keys upon login. HTH, /p Quoting Jacek Generowicz (2012-04-03 16:29:07) Hello, I'm trying to use notmuch from Emacs. I have a number of SMTP accounts through which I would need to send mail. Suggested solutions I have found fall into 3 categories: + Gnus posting styles. (I'd rather stay clear of gnus, at present.) + Writing plaintext passwords in the MSMTP config files. (I'm rather relutctant to leave plaintext passwords lying around.) + Emacs' own SMTP library. (I don't understand how to get it to deal with many accounts in a non-gnus based way. It also seems to be undergoing an interface change in the transition from Emacs 23 to 24, and it was already confusing enough without that complication.) To further complicate matters, I'm looking for a solution which queues any messages which are sent while the machine is offline. (It seems that MSMTP and Emacs SMTP library do offer some help in those respects.) Hints gratefully received. Thank you. ___ notmuch mailing list notmuch@notmuchmail.org http://notmuchmail.org/mailman/listinfo/notmuch ___ notmuch mailing list notmuch@notmuchmail.org http://notmuchmail.org/mailman/listinfo/notmuch
Re: Passwordless multi-account SMTP
On Tue, 3 Apr 2012 17:29:07 +0200, Jacek Generowicz jacek.generow...@cern.ch wrote: Hello, I'm trying to use notmuch from Emacs. I have a number of SMTP accounts through which I would need to send mail. Suggested solutions I have found fall into 3 categories: + Gnus posting styles. (I'd rather stay clear of gnus, at present.) + Writing plaintext passwords in the MSMTP config files. (I'm rather relutctant to leave plaintext passwords lying around.) + Emacs' own SMTP library. (I don't understand how to get it to deal with many accounts in a non-gnus based way. It also seems to be undergoing an interface change in the transition from Emacs 23 to 24, and it was already confusing enough without that complication.) To further complicate matters, I'm looking for a solution which queues any messages which are sent while the machine is offline. (It seems that MSMTP and Emacs SMTP library do offer some help in those respects.) Hints gratefully received. I use a slightly modified msmtpq for queuing and I have some hooks that run profiles to choose SMTP server and other options based on sender/receiver/last search. Passwords are plain text but my fs is encrypted. If you want, you could encrypt the password file with gpg and use gpg-agent to grab the password as needed from the file. Good luck. pgph4o5PyLp8d.pgp Description: PGP signature ___ notmuch mailing list notmuch@notmuchmail.org http://notmuchmail.org/mailman/listinfo/notmuch
Re: notmuch-poll and OfflineIMAP
Jacek Generowicz jacek.generow...@cern.ch writes: Hello, Would you have any advice on how to construct a notmuch-poll script that would work well in concert with OfflineIMAP? In particlular, how can you avoid having to re-issue the IMAP account passwords? The OfflineIMAP docs state that SIGUSR1 forces an immediate resync of all accounts, so something based around kill -SIGUSR1 `cat ~/.offlineimap/pid` or pkill -SIGUSR1 -u `whoami` offlineimap (with an already-running OfflineIMAP process in autorefresh mode) could work, but it's not clear to me how to discover when the resync is done, and sleeping for an arbitrary time before calling 'notmuch new' seems unsatisfactory. Any words of wisdom? Offlineimap supports running hooks after syncing. Check if SIGUSR1 does that too. Then you could run notmuch new in the post sync hook. Also, notmuch new supports pre-new and post-new hooks. You could start or SIGUSR1 offlineimap in the pre-new hook. See man notmuch-hooks for details. HTH, Jani. ___ notmuch mailing list notmuch@notmuchmail.org http://notmuchmail.org/mailman/listinfo/notmuch
Re: Handling of symlinked maildirs?
On Mon, 2 Apr 2012 10:54:48 -0700, Jameson Graef Rollins jroll...@finestructure.net wrote: On Mon, Apr 02 2012, Andrei POPESCU andreimpope...@gmail.com wrote: Im sorting my mailing lists with generic maildrop rules like this one: if (/^List-Id:.*debian-(.*)\.lists.debian.org/) to Maildir/.debian.$MATCH1/ However, I'm subscribed to a *lot* of mailing lists and in order to keep my folder view sane I use symlinks to conflate some of them, e.g. .debian.devel-announce - .debian.devel This works well since mutt simply ignores the symlink(s) so I don't even need to exclude them in the config, but it seems that notmuch does index each of the symlinks as a separate folder[1]. Does it make sense to have this configurable or even completely exclude the symlinks? Hey, Andrei. I would say that if you have a compelling reason that notmuch should be able to ignore symlinked directories then it maybe makes sense to have it configurable. However, I must say that I have completely dropped the whole folder concept myself since I find it totally redundant with notmuch's tagging and searching capabilities. Please don't exclude the indexing of symlinks in notmuch. I have a fairly restricted IT environment at work, where I have chosen to symlink all my maildirs to other disks so that they appear in my home directory but do not fill my home disk quota every day. Often my different maildirs are to make my non-notmuch mail viewing slightly tolerable. There was a configuration option (new.ignore) added to 0.12 that lets one specify names of things not to index. That may be enough to do what you want, although not as automatic as just pitching all symlinks while indexing. Thanks, -Mark ___ notmuch mailing list notmuch@notmuchmail.org http://notmuchmail.org/mailman/listinfo/notmuch
Re: notmuch-poll and OfflineIMAP
Hi Jacek, On Tue, Apr 3, 2012 at 09:27, Jacek Generowicz jacek.generow...@cern.ch wrote: Would you have any advice on how to construct a notmuch-poll script that would work well in concert with OfflineIMAP? In particlular, how can you avoid having to re-issue the IMAP account passwords? The OfflineIMAP docs state that SIGUSR1 forces an immediate resync of all accounts, so something based around kill -SIGUSR1 `cat ~/.offlineimap/pid` or pkill -SIGUSR1 -u `whoami` offlineimap (with an already-running OfflineIMAP process in autorefresh mode) could work, but it's not clear to me how to discover when the resync is done, and sleeping for an arbitrary time before calling 'notmuch new' seems unsatisfactory. Any words of wisdom? I run offlineimap in autorefresh mode (with IDLE too, actually) and use a script based on inotify to update notmuch whenever new mail comes in: https://gist.github.com/1952483 . It sounds like some variant of this would probably work for you. As an added bonus, I have my post-new script run emacsclient -n -e '(notmuch-hello-update t)', so if I flip to my mail desktop I can see right away if there's new mail. ___ notmuch mailing list notmuch@notmuchmail.org http://notmuchmail.org/mailman/listinfo/notmuch
Re: [PATCH 6/8] cli: add support for batch tagging operations to notmuch tag
On Tue, Apr 03 2012, Jani Nikula j...@nikula.org wrote: Yes, it's more flexible, but what are the real benefits of such flexibility? What other commands than tag/restore would truly benefit from this? I might add that nobody has asked for such flexibility. But I do believe it was your patch that introduced the idea of batch processing of other commands to begin with. I think you're just turning my original question to you back on me. There's no point in adding and initial flag to the beginning of the line to specify the command to execute if you have no intention of ever supporting other commands. Or vice versa, if you ultimately want to support multiple commands then the place to do it is in the top level interface, and not in one of the subcommands. Ultimately, I think this patch series suffers from being simultaneously too close to existing functionality (restore), yet tantalizingly close to a much more generally useful concept of batch command processing. I would rather see it go one way or the other rather than add something fairly redundant in the middle. jamie. pgpwiucwk4x3v.pgp Description: PGP signature ___ notmuch mailing list notmuch@notmuchmail.org http://notmuchmail.org/mailman/listinfo/notmuch
Re: Message deletion wisdom
On Tue, Apr 03 2012, Jacek Generowicz jacek.generow...@cern.ch wrote: Looking through the archives of this list, I see that the topic of message deletion is, if not controversial, then at least non-trivial. + Requests for the provision of message deletion mechanisms have appeared many times, and that various patches addressing the issue have been submitted, and (IIUC) some of these are awaiting review and might possibly make it into an official release in the future. + Having the 'deleted' tag cause messages to be deleted is, apparently, fraught with peril of losing mail. Hi, Jacek. You are right on both of these points. It has indeed been a controversial topic. However, I would say that we have reached a stasis in terms of our response to this issue. I'll try to summarize here. I can say with a very high degree of certainly that notmuch will NEVER gain the capability to actually delete mail files from disk itself. It is too risky for notmuch to be involved in that, and it's too easy to do it outside of notmuch (e.g. notmuch search --output=files tag:deleted | xargs rm). This has never really been up for discussion. What *has* generated a lot of discussion is whether or not the emacs interface should by default support key bindings that add a deleted tag to messages, and how that tag should be handled. I think we have finally settled on notmuch NOT supporting such a key binding by default, since it's easy enough for users to add the binding themselves if they wish. What we have added (or more accurately are adding) though (thanks to the work of Austin Clements and Mark Walters) is the ability to exclude messages with a certain tag from search results. This is like an effective deletion, since messages can be completely omitted from search results without having them actually be deleted from disk. This is obviously much safer, and a more generally useful functionality. This functionality is currently in beta, but will be full-featured in the next release. So in summary, if you would like to delete messages, you can: * add a key binding to emacs (or your favorite ui) to add a deleted tag to messages that you want to delete: (define-key notmuch-show-mode-map d (lambda () (interactive) (notmuch-show-tag-message +deleted))) * add exclusion for messages with the deleted tag: notmuch config set search.exclude_tags deleted * and if you really want them purged from disk, delete them manually with: notmuch search --output=files tag:deleted | xargs -l rm I'll add a page about this to the wiki, since it comes up enough. hth. jamie. pgp0FHgBRTDz9.pgp Description: PGP signature ___ notmuch mailing list notmuch@notmuchmail.org http://notmuchmail.org/mailman/listinfo/notmuch
Re: [PATCH 6/8] cli: add support for batch tagging operations to notmuch tag
Jameson Graef Rollins jroll...@finestructure.net writes: There's no point in adding and initial flag to the beginning of the line to specify the command to execute if you have no intention of ever supporting other commands. My thinking was that it was useful for the disk format to have a bit more information in it so that we could more easily change the interface in an upwardly compatible way. If at some point in the future we do have more general batch command processing, it would be nice not have to change the file format again, particularly for dump files. Ultimately, I think this patch series suffers from being simultaneously too close to existing functionality (restore), Well, as Jani mentioned there is a fair amount of code overlap with an in-progress improved dump/restore facility. yet tantalizingly close to a much more generally useful concept of batch command processing. Well, opinions seem to differ about how close we are to that in terms of real code, but I think it is worth thinking about as we define file formats. d ___ notmuch mailing list notmuch@notmuchmail.org http://notmuchmail.org/mailman/listinfo/notmuch