Re: mairix search
On 03.05.11,00:26, Tim Gray wrote: On Apr 29, 2011 at 01:56 PM +0200, Sebastian Tramp wrote: Is searching / indexing with mairix state of the art or is there a better solution available? I am quite happy with that, just wanted to ask ;-). I found that mairix was a lot better for me than nmzmail. I used mairix for quite a bit. I then found mu (and figured out how to compile it on OS X). I liked that even more. Now however, I'm running notmuch as just an indexer. It's pretty fast. I really like it. It's also being actively developed. And the best part in my mind is that it's developed as a library with a command line utility. So my secret dream is that some enterprising developer hooks notmuch directly into mutt. That would be awesome. I use mairix, but it seems like mu is being quite actively developed: http://www.djcbsoftware.nl/code/mu/ Jostein
Re: mairix search
On Tue, May 03, 2011 at 12:26:45AM -0400, Tim Gray wrote: On Apr 29, 2011 at 01:56 PM +0200, Sebastian Tramp wrote: Is searching / indexing with mairix state of the art or is there a better solution available? I am quite happy with that, just wanted to ask ;-). I found that mairix was a lot better for me than nmzmail. I used mairix for quite a bit. I then found mu (and figured out how to compile it on OS X). I liked that even more. Now however, I'm running notmuch as just an indexer. It's pretty fast. I really like it. It's also being actively developed. And the best part in my mind is that it's developed as a library with a command line utility. So my secret dream is that some enterprising developer hooks notmuch directly into mutt. That would be awesome. Sounds interesting. do you think, notmuch is faster than mairix (not that I have a problem with mairix speed, just for our information) Can you share your config parts regarding notmuch / mutt integration? (or do you use it via emacs?) best regards Sebastian Tramp -- Sebastian Tramp WebID: http://sebastian.tramp.name
Re: mairix search
On Mon, May 02, 2011 at 06:00:04PM -0400, Ed Blackman wrote: On Fri, Apr 29, 2011 at 02:08:33PM -0700, Omen Wild wrote: Quoting Ed Blackman e...@edgewood.to on Fri, Apr 29 17:03: In my case, mairixquery is a Perl script that prompts me for the mairix search string, gives me yes or no prompts for whether to search threads or augment previous results, and saves the most recent 100 searches so that I can make edits if the query returns too little or way too much. That sounds really interesting. Any chance you would be willing to share it? Sure. I've attached it. great, it works for me. just a notice for the other: under debian/ubuntu, there is a dependency for the libterm-readline-gnu-perl package ... best regards Sebastian Tramp
Re: offlineimap much slower than gmail-imap
On Mon, May 02, 2011 at 12:03:46AM +0200, Jose M Vidal wrote: I was happily using mutt with gmail-imap. Then I decided to switch to offlineimap+msmtp, so I could still use mutt offline, have a backup of all my e-mails and, hopefully, increase mutt speed by working locally. But, after having all installed an all my e-mails already downloaded (10GB / 45.000 e-mails), every time I switch from INBOX to All Mail, it takes 6 minutes (!) to refresh the index. I had caché enabled with imap, and I am stilll keeping it in my .muttrc: set header_cache=~/.mutt/GMail/cache/headers¬ set message_cachedir=~/.mutt/GMail/cache/bodies¬ But I am afraid cache only works with online imap, because now accessing to folders is much-much slower than with previous online configuration. Is there anything I can do to speed-up my mutt? Thanks in advanced, -- jm PS: my mutt is 1.5.20 If your maildir is on an ecryptfs filesystem, the problem is the stat() call that mutt makes to check for changes in the maildir files. This is very slow on an ecryptfs filesystem because the header on every file must be decrypted for the stat() call. If you are certain that no other program might modify your mail files, you can unset maildir_header_cache_verify to bypass the stat() call. I use this for my large archive folders (which I also set read-only for safety) with a folder hook: folder-hook 'archive' 'push toggle-write; unset maildir_header_cache_verify' After I did this, access to my large folders became virtually instant. -- Scott Barker sc...@mostlylinux.ca Linux Consultant http://www.mostlylinux.ca/scott
Re: mairix search
On May 03, 2011 at 08:39 AM +0200, Jostein Berntsen wrote: I use mairix, but it seems like mu is being quite actively developed: Yes, mu is quite actively developed. I liked it a fair amount. I just have a feeling that notmuch has a brighter future.
Re: offlineimap much slower than gmail-imap
Hi everybody. Thank you very much for your help. The situation now is: 1- As my /home folder is ecrypt I added to my .muttrc your suggestion (folder-hook 'archive' 'push toggle-write; unset maildir_header_cache_verify') Apparently, after a first refresh, the update of files looks inmediate, but after using mutt a little bit, it turns slower againg. The end result is that there's not a significant improve in performance: Sometimes it is very fast, sometimes it is very slow. 2- I checked my data, ang got this: jm@jm-ThinkPad-X200s:~/.mail/GMail/[Gmail].All Mail$ du -h 4,0K./tmp 940K./new 9,4G./cur 9,4G. jm@jm-ThinkPad-X200s:~/.mail/GMail/[Gmail].All Mail$ find . -type f | wc -l 47860 The data shown in my mutt status page is the same (I am using google apps, premium, so nothing strange about having more than 7Gb) 3- My system is 32 bit. jm@jm-ThinkPad-X200s:~$ uname -m i686 4- After working a little bit with I/O, I get: TID PRIO USER DISK READ DISK WRITE SWAPIN IOCOMMAND 120 4886 be/4 jm 2.38 M/s0.00 B/s 0.00 % 96.43 % mutt Honestly speaking, I don't know what to do with this information. Thanks again for your help. jm PS: will keep tokyo-cabinet for later, thanks. Think that understanding what is happenings is neccessary first.
Re: mairix search
On May 03, 2011 at 08:47 AM +0200, Sebastian Tramp wrote: Sounds interesting. do you think, notmuch is faster than mairix (not that I have a problem with mairix speed, just for our information) It's been a while since I've used mairix. I forget exactly how long it took to index things initially. I do remember however that I would only reindex to look for new mail once or twice a day because it took a fair amount of time and cpu power. I'm currently indexing about 260,000 messages across many maildirs. For unimportant reasons, I just reindexed them all from scratch last night. It took 1 hour and 10 minutes. I did this while I was video chatting a friend. This was an 'initial' index. Subsequent indexing, looking for newly delivered messages, takes between 20-40 seconds, so I can run it every time I check for mail. Which I do - I have it in my getmail script. You can run it any time you want with the command `notmuch new`. The other thing that I liked about notmuch infinitely more than mairix, and a fair bit more than mu, is that the search syntax just feels more natural to me. Your mileage may vary. Can you share your config parts regarding notmuch / mutt integration? (or do you use it via emacs?) I don't use it via emacs. The mutt integration is very similar to what you'd do with mairix or mu. A couple of bindings that just run the command line 'notmuch' program with your search terms following. You do need to do a bit of command line piping to turn the filenames that notmuch outputs as search results into symbolic links. Then you jump to a maildir with the results. Though you can run it directly like this, I wrote a little python script that I call instead which gives me readline search history and cleans out the search results automatically when I make a new search. I've attached the script 'notmuch-mutt.py'. I started building in support for notmuch's tagging features but kind of lost interest. You could also write an option if you want to *not* clear out the previous search results; then you could build up results incrementally if you desired. I have a couple bindings defined. I'll try to list them below; hopefully they don't get too jumbled up since they are kind of long (I don't break up my bindings definitions into separate lines in my muttrc - hopefully the \'s I put in break things appropriately). This one just changes to the search results folder: macro index,pager .r change-folder-readonly \ ~/.notmuchmutt/search enter This on runs a search: macro index,pager .s enter-commandunset wait_keyenter \ shell-escape~/bin/notmuch-mutt.py -penter \ change-folder-readonly~/.notmuchmutt/searchenter \ enter-commandset wait_keyenter This on runs a search and includes the full thread in the results. You can also run it on a search result to reconstruct the thread from the message of interest. macro index,pager .t enter-commandunset wait_keyenter \ pipe-message~/bin/notmuch-mutt.py - --threadenter \ change-folder-readonly~/.notmuchmutt/searchenter \ enter-commandset wait_keyenter This runs muttjump on the current message (see below). I don't do all the screen stuff that muttjump can do. macro generic .o enter-commandpush pipe-messagemuttjump \ enterenter jump to original message notmuch also has python bindings, so you can access the library directly. For some reason on OS X, they don't work right for me, so my python script just calls the command line utility directly. I also use the awesome 'muttjump' script [1]. It takes you to the parent mailbox of a message in the search results. Hopefully I didn't forget any important bits. The actual notmuch config file is very simple as there aren't a lot of relevant options. You just set the directory to your maildir, and define your user name and address (I think this is used only if you are running it in emacs or vim as a full mail client). Lastly, you can set which tags get set for new messages and whether or not you want imap flags (read, flagged, etc.) that are set in notmuch synchronized back to the maildir files. Again, the last option is mostly for the full notmuch client. [1]: https://github.com/weisslj/muttjump #!/usr/bin/env python __author__ = Tim Gray __version__ = 1.0 import sys import os import optparse import subprocess as sb import shlex import email import readline cfgdir = '~/.notmuchmutt' cfgdir = os.path.expanduser(cfgdir) searchdir = os.path.join(cfgdir, search/) searchhist = os.path.join(cfgdir, 'search-history') taghist = os.path.join(cfgdir, 'tag-history') if os.path.isdir(cfgdir): cfgFlag = True else: print 'must run with --config first to set up directory' sys.exit(10) class notmuch(): def __init__(self): self.nm = '/usr/local/bin/notmuch' def runCmd(self, query): Runs the given command. cmd
Re: offlineimap much slower than gmail-imap
What does the following command give you (assuming your disk is /dev/sda): hdparm -tT /dev/sda jm@jm-ThinkPad-X200s:~$ sudo hdparm -tT /dev/sda6 /dev/sda6: Timing cached reads: 2738 MB in 2.00 seconds = 1370.52 MB/sec Timing buffered disk reads: 116 MB in 3.02 seconds = 38.45 MB/sec PS: note I had to sudo it. My /home doesn't have a specific partition this time: jm@jm-ThinkPad-X200s:~$ df -h S.archivosTam. Usado Disp. % Uso Montado en /dev/sda6 45G 29G 14G 69% / none 950M 320K 950M 1% /dev none 956M 1,7M 954M 1% /dev/shm none 956M 100K 955M 1% /var/run none 956M 0 956M 0% /var/lock /home/jm/.Private 45G 29G 14G 69% /home/jm Thanks again, -- jm
quoted-printable not displaying correctly
I've noticed that pretty much all of the e-mail I get from MS Windows users is not being displayed correctly. The problem messages are multipart/alternative. There are two parts: The first is Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable The second is Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable The latter gets displayed properly when filtered through w3m, but I'd rather not to do that, so I have mutt configured to prefer the text/plain alternative, but it isn't getting displayed properly. The text/plain body looks something like this: Weird... where did all those question marks come from?=C2=A0 I just checked= my sent file and they're not there.=C2=A0 I'm gramatically correct.=C2= =A0 -jill=20 And it gets displayed like this: Weird... where did all those question marks come from??? I just checked my sent file and they're not there.?? I'm gramatically correct.?? -jill As you can see, the =C20=A0 sequences get displayed as ??. The =20 sequences and the end-of-line = characters seem to be handled OK. How do I fix this? -- Grant Edwards grant.b.edwardsYow! I am NOT a nut at gmail.com
Re: quoted-printable not displaying correctly
On 2011-05-03, Grant Edwards grant.b.edwa...@gmail.com wrote: I've noticed that pretty much all of the e-mail I get from MS Windows users is not being displayed correctly. The problem messages are multipart/alternative. There are two parts: The first is Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable I didn't have LANG (or any of the LC_ variables) set. Setting LANG to en_US.UTF-8 and running mutt in a UTF8 capable terminal fixed the problem. -- Grant Edwards grant.b.edwardsYow! Here we are in America at ... when do we collect gmail.comunemployment?
