Strigidaemon

[Robert Knight]

Hi,

> It should not be set to run everytime a computer is started up.  An 
> application like strigi needs to be very much under the control of the user, 
> for example:

In all fairness, it isn't an easy problem to solve but competitors on other platforms (Windows Vista's Indexer 
and MacOS' spotlight) make a pretty good effort to index everything automatically with no configuration
and minimal CPU usage.  Tracker managed quite well last I tried it though I have seen a fair 
few complaints as well.

>From the presentation Jos gave at Akademy, the primary goal of Strigi was to index a lot of data
very fast, relying on the kernel's scheduler to balance the needs of the Strigi indexer against other
programs.  This is okay if you have a specific directory full of non-text files which you know you want 
to search and can manually index only when needed.  Strigi then becomes a highly capable replacement for locate/updatedb.
If you want a general desktop search solution that everybody can use then the requirements are quite different.  
The indexer doesn't have to be really fast but it does need to minimize processor usage, especially in
laptops running on battery and it has to be very robust with respect to corrupt files, buggy plugins etc.

> a configuration 
> screen should come up (and allow configuration to be deferred and "remind me 
> later"). 

Please no.  This is 2008 not 1990, I don't expect to have to do a whole
load of manual configuration to search my email and lecture notes.  If
Strigi isn't intended as a general purpose search tool then there isn't
a lot of need to tell users about it until they have a folder of files
that they want to search.

Regards,
Robert.

On Fri, 2008-03-28 at 13:12 -0400, Randy Kramer wrote:
> On Friday 28 March 2008 12:36 pm, Anders Lund wrote:
> > On fredag 28 Marts 2008, Jos van den Oever wrote:
> > > 2008/3/28, Anders Lund <anders at alweb.dk>:
> > > > What is wrong with strigidaemon?
> > > >
> > > >  It can't be that it is supposed to run at 80-99%CPU for hours after i
> > > > boot! That is simply not acceptable.
> > > >
> > > >  It is set to index my home directory.
> > >
> > > It is well possible if you have a lot of data. Hours is however a very
> > > long time.
> > 
> > I have quite a lot of data, and for example kde source is quite a few files. 
> I 
> > still think it's overly long time, and I don't understand why it should run 
> > everytime i start up.
> 
> It should not be set to run everytime a computer is started up.  An 
> application like strigli needs to be very much under the control of the user, 
> for example:
> 
> The first time the computer starts (with a new distro or kde), a configuration 
> screen should come up (and allow configuration to be deferred and "remind me 
> later").  On that configuration screen it should be easy to select which 
> directories to be indexed (allowing easy exclusion of selected 
> subdirectories), and when they are to be indexed (again, easily allow 
> different directories to be indexed at different times.
> 
> The selections should also allow the user to specify which files to index, by 
> extension and/or MIME type.
> 
> The selections of when to index should include options like:
>    * at specified times (which will end up most likely being nights) and dates 
> (I wouldn't index all my files every night, for one thing it's unnecessary 
> (sp?) wear and tear on the disk drive, as well as unnecessary power use--I 
> might index a certain small set of files daily, others I might index monthly 
> or less often (or maybe just on demand)
>    * just before shutdown (i.e., an option to start indexing when the user 
> calls for a shutdown, and then shutting off the computer when indexing is 
> complete)
>    * at startup--I'm almost willing to bet no one will choose this unless you 
> make it the default choice, in which case see my last paragraph, below
> 
> In addition, there should be easy to use tools on the toolbar to do things 
> like:
>    * index a specified directory now
>    * temporarily interrupt (i.e., pause) any indexing in progress (and then a 
> similar means to resume  (not restart, but with an option to restart) that 
> indexing)
> 
> Anything less, and you're just asking for a ton of user complaints.
> 
> Randy Kramer
>