Mailinglist Archive: opensuse (880 mails)

< Previous Next >
Re: [opensuse] KMail on 12.3
  • From: "Carlos E. R." <robin.listas@xxxxxxxxxxxxxx>
  • Date: Tue, 27 Aug 2013 15:24:53 +0200 (CEST)
  • Message-id: <alpine.LNX.2.00.1308271503220.9215@minas-tirith.valinor>
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256



El 2013-08-26 a las 19:37 -0400, Anton Aylward escribió:
Carlos E. R. said the following on 08/26/2013 06:15 PM:


So I while I may not have understood you in some aspect (as below) what we are saying about this part is much the same.

Yes :-)


On a maildir folder, it just writes 2 or 3 small files, it is faster, in
theory.

Ah.
My INBOX is about 152Meg (according to ls -h) and I have no complaints about the performance of Dovecot handling it. Theory is good. Perhaps this would be different is this was not a personal setup, of that was a larger server with tends of gigs or RAM, a couple of 8 core processors and few hundred T of disk at an ISP support a few thousand concurrent users with webmail, each with INBOXes they never purged and which were many times the size of yours and mine.

But then you have the issue of a filesystem with billions of small files...

There are mail folder formats intermediate between maildir and mbox. Say that small mails are stored together in the same file till it grows to a few megabytes, and large emails are stored alone on a single file. In theory it should work very well.

Pine (it happens that one of the developers of pine was the designer of Imap, so what pine does is almost canonical, it follows standards to the letter) started developping a new such format, but was not implemented finally, or I don't know how to use it. And it is pointless if procmail does not support it.

Dovecot (specially the new release) supports such formats, too.

But then we can't use procmail because it does not support them.



I'm not sure about that. Personally, without looking at the source, I think the MUAs do the sorting at the presentation.

True, but Pine can also write a sorted folder. Thunderbird does not, but
thunderbird keeps an index file (and pine does not).

Can pine deal with threads then?

Oh, absolutely. Default sort is none (last post arrived is at the bottom). I prefer threaded sort with threads with recent posts at the bottom (or top).


I wish T'bird didn't; I wish it could make use of the capabilities of Dovecot in that regard. I wish it could use dovecot's ability to use lucine for full text indexing too and allow easy body search.

I wish I could win the lottery ...

:-)


I don't know if Pine uses server index or local.


Thunderbird is designed mainly with ISP imap as a target, where a local index is fast. Pine doesn't create indexes except perhaps in memory, and it assumes a good network (the local network of the university where it was designed, I guess).


But I meant adding a header inside mails in the mbox file, not in an
index. There was a client that did it (I can't remember the name, the
header might be named "X-COMMENT" or similar. Addition of headers is
permitted.

It shouldn't make any difference. I can easily imagine a mail system implemented not as files but as a database where the headers are stored in different fields from the body and its easy to add another header.

So what difference does it make if the metadata isn't stored with the data? The same index file that points to the message could equally well contain any number of X- extensions without including them in the message.

Heck, news readers manage that; they keep track of who has read a message without making changes to the message.


Altering a header in the message, instead of using the index file, means that the index can be rebuilt, without losing that data. Many years ago I started designing a data base, and that was a feature: I could delete the index and rebuild them safely and reliably.

In any case, a header is "data", not "metadata", in mail context.

Yes, mail is a database, only that it is a specific one, and everybody redesigns it again.


Absolutely, but feeding it via dovecot tools would avoid having to reindex
the entire folder.

Yes, so?
That is only of use if you can access and make use of the dovecot index.
Most MUAs can't. I'm annoyed that Thunderbird can't.

I'm not sure about Pine.

It seems pointless to me that I should run Dovecot on a server and let
it do indexing and full text indexing "in the background" with the
intent to offload this processing and storage from my workstation and/or laptop. It seems pointless to me that I should have T'bird index things all over again.

TH: Edit/Find/Search messages. There is a tick box for "run search in server". Is that what you want? :-)


Mind, there are two types of indexes here.

There is one, telling just some headers used for sorting and displaying headers, and then there is the contend index used for searches. Possibly Th uses the first one, not the second.


Yes, I know about
https://wiki.mozilla.org/Thunderbird:Using_Gloda
and I have found information on that less than useful and experimentation frustrating.
This
https://developer.mozilla.org/en-US/docs/Thunderbird/Gloda_indexing
leads me to believe that in IMAP mode T'bird does a lot of indexing that would be redundant of it could access the dovecot metadata.


There is
https://addons.mozilla.org/en-us/thunderbird/addon/glodaquilla-search-indexing-en/
but is it what we are talking about? I don't think so.

I'm not at home, my internet is limited now to 500MB/month. I'll look at those later.

- -- Cheers
Carlos E. R.

(from 11.4, with Evergreen, x86_64 "Celadon" (Minas Tirith))

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.16 (GNU/Linux)

iF4EAREIAAYFAlIcqKYACgkQja8UbcUWM1yfeAD/Wo7Au7eozWTDnr8ckgXj0TtK
2k9a2HVKpfUkCVKtqxcA/0lHtn/T8mICJLvu7PpItQdAqQl/SrIOuOhFJbTbKGZZ
=/MfO
-----END PGP SIGNATURE-----
< Previous Next >
Follow Ups