[Tracker] [Fwd: Re: strigi] Strigi-Tracker colaboration

[Ivan Frade <ivan frade nokia com>]

Hi all,

 Last weekend i was in Guademy (http://2008.guademy.org). I presented
there tracker and what we are doing on it, and our short term plans. I
also talked with some KDE guys and i saw a very good disposition to
collaborate between desktops.

 So i sent a message to the strigi-devel mailing list and received some
interesant feedback. My last response is attached to this mail. Comments
are welcome. What do you think?

 Of course, we are not talking about merge both projects. Each one has
his own style and it is great. But it makes no sense to reimplement the
same things again and again.

Ivan

--- Begin Message ---

• From: Ivan Frade <ivan frade nokia com>
• To: ext Jos van den Oever <jvdoever gmail com>
• Cc: Strigi-devel lists sourceforge net
• Subject: Re: strigi
• Date: Tue, 29 Apr 2008 14:13:35 +0300

hi jos,

El lun, 28-04-2008 a las 22:59 +0200, ext Jos van den Oever escribiÃ:

2008/4/28 Ivan Frade <ivan frade nokia com>:

Hi Ivan,

Nice to hear from you. I like that Nokia is looking at Xesam. It will
help make it a good standard. 

 I think there is a rough consensus around the specification, and it is
very good. It can be also pretty fun: all projects can implement the
XESAM interface and then let the user choose the faster/lighter/ search
engine.

Areas in which we can offer code for use in Tracker is the library
libstreamanalyzer. It does all our heavy lifting with respect to
parsing and indexing files. It is written in C++ and i'm sure that's
an obstacle though.

 Uhmmm Does libstreamanalyzer do something similar to our
"tracker-extractor"? 

 If it is so, could we implement (for example) tracker-extractor using
libstreamanalyzer? It sounds like a good exercise. It would be an
external program so the programming language is not an issue.

 Then we could test the performance and how easy is to write new
extractors/analyzers.

 By the way, what license is using libstreamanalyzer? What license must
be used for new analyzers? 

Another thing we could work on together is on having a shared lib or
shared daemon for notification of changes in files.

 After the refactoring, we expect to have a quite well isolated crawler
module. I never though about split it in a library but it sounds
reasonable from a design point of view (let see what happens with the
performance). Strigi can get a stable implementation of inotify and both
Tracker and Strigi can work together implementing the windows/other
systems support.

I'd love to hear your ideas for collaboration.

 Some other ideas to think about:

* A common set of data to test performance? It could be part of the
XESAM specification, besides the current tests. It is a lot of work to
prepare a good set, so we can share the effort.

* A common configuration system (offering in DBus the configuration
object?) for the indexer. We need to configure the same: directories to
crawl, pause in the indexing, language, steeming yes/no, full text
yes/no...

 Our plan is to release a 0.7.0 version of tracker with the refactored
code and the initial XESAM implementation. It will take some time, but
then we will have a really nice and well organized code.

 Regards,

Ivan

--- End Message ---