Re: [Tracker] Question about the Roadmap



Le mardi 21 novembre 2006 Ã 17:22 +0000, Jamie McCracken a Ãcrit :
Laurent Aguerreche wrote:
Le mardi 21 novembre 2006 Ã 17:26 +0100, Javier Arantegui a Ãcrit :
Hello,

It seems that Tracker is getting better very rapidly. I've read its
roadmap (*), and I wonder if its development is following it. I mean,
I can wait until January when it's planned that Tracker will index
e-mails (my most needed feature), but I would like to know if the plan
is followed or, even better, if the email indexing could be move
forward.

I am currently working on email indexing and I expect to send a patch
which will make Tracker store emails in its database by the end of the
week. But it will not let you query database to retrieve your emails
(for the moment).


It would be best not to do that.

I will be changing the DB structure in next release to make way for our 
new more powerful metadata system. Currently we can not handle metadata 
that has multiple values for the same type (such as Email.To which can 
contain many addresses) but the new system will enable us to overcome 
these limitations.

Instead of tryting to save them in the DB, there are more urgent fixes 
needed:

The big fixes that are needed for Evolution emails:

1) Get correct URI - currently its wrong or shows the wrong email (I 
expect this is to do with the UID  being in hex and not decimal - not 
sure which the uri uses?)

I found why... The X-Evolution field is written that way :
  X-Evolution: uid-flags

In my code, I inverted respective places of UID and Flags! :-)
There were also a problem about the format used with
g_strdup_printf()...

I can send the patch...

2) Need to dynamically index all mbox files - looking at beagle's source 
this means watching all files with .ev-summary ending (only mboxes have 
a corresponding summary file EG Sent.ev-summary  ). Currently its hard 
coded to Sent and Inbox which is useless for virtual folders and does 
not pick up my Gnome mbox (gnome and gnome.ev-summary is there).

The patch I was speaking about also uses .ev-summary files to find
MBox.  :-)


3) Evolution caches remote emails from an smtp server in a different 
place and with different extensions/summary files - we will need to 
crack those (Beagle source again is useful here)

I has not sent the patch due to that... I do not have this cache. I read
code in libcamel (it is used by Evolution to store/read email +
write/read ev-summary and summary files) and I only found code about
that cache into evolution-data-server-1.8.1. On my computer, I only have
version 1.6 so I think I have to update some packages from Debian
Experimental.

But, is there somebody who uses 1.8 and who could run :
   find ~/.evolution -name "*-meta"
??
I would like to be sure that the code in libcamel is used!


This cache is quite interesting because it only contains:
   timestamp, UID, flags, offset (in the file)
for each email.
So when an existing email is modified (placed in junk for instance) I
can parse the cache quickly to find it instead of parsing MBox file
(ouch!) or .ev-summary file (a little bit bigger than cache).


I will be happy to do the DB end of things for the emails once the new 
metadata system is ready but lets get most of the above fixed first :)


Laurent.

Attachment: signature.asc
Description: Ceci est une partie de message =?ISO-8859-1?Q?num=E9riquement?= =?ISO-8859-1?Q?_sign=E9e?=



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]