[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-devel
Subject:    Re: Access indexing
From:       Ian Wadham <iandw.au () gmail ! com>
Date:       2012-02-25 10:44:10
Message-ID: 58B2812C-77E9-44E3-82F6-88EB9AB12771 () gmail ! com
[Download RAW message or body]

On 25/02/2012, at 8:14 PM, Stephan Menzel wrote:
> On Sat, Feb 25, 2012 at 12:44 AM, Ian Wadham <iandw.au@gmail.com> wrote:
>> Yes I do (have a Macbook).  The search function is a small magnifying
>> glass icon at the top right of the screen (equivalent to a "systray" item).
>> It drops down to single line-edit type field with the label "Spotlight".
> <snip>
> It takes some time and change in behaviour but if you really give it
> a shot it's a good thing indeed. But as you said, most people really
> try that only once. If it doesn't do as expected the first times, they
> will simply deactivate it and stick to the old ways. Semantic Desktop
> nil, dusty old ways score.  So I think accessibility and simplicity is
> key here and has to be done right.

Well I do use Google and Wikipedia a lot.  Also, on my iPhone, I use the
textual search in the Maps app quite a bit.  However I do not have frequent
need for a search engine on my desktop.

That is despite having pioneered the use of text-search engines in the
Australian Government in the early 1980s, including indexing on fields
and attributes in searches ("Brown" the name versus "brown" the colour)
and indexing on terms from a hierarchical thesaurus, which I think are
basically what Nepomuk and Strigi are trying to do, even if they are
cloaked in much obscure talk about "ontologies".

>> I am surprised too that Nepomuk, Strigi et al. require so much CPU
>> power and space.  This is not at all what was claimed at Akademy
>> 2007, akademy2007.kde.org/conference/slides/strigi.pdf  See the
>> slide entitled "Speed Comparison".  What went wrong?
> 
> Well, I have to correct my initial rant about that a little.

I was impressed.  It had all the characteristics of a truck hitting a sacred cow!

> There was
> one problem here I could fix. During the upgrade from 4.7 to 4.8 there
> are some temporary files left in /tmp. When you start 4.8 and those
> files are still there, it causes problems in virtuoso-t. So I stopped
> KDE, deleted everything in /tmp and restarted KDE. This caused some
> initial reindexing and resource demand but it actually stopped after a
> while and isn't present all the time anymore.
> Most of the time though it still is
> 
> top - 10:11:52 up 16:34,  4 users,  load average: 0,67, 0,69, 1,02
> Tasks: 130 total,   2 running, 128 sleeping,   0 stopped,   0 zombie
> %Cpu(s):  3,4 us,  8,3 sy, 61,4 ni, 26,1 id,  0,9 wa,  0,0 hi,  0,0 si,  0,0 st
> Kb Mem:   1919532 total,  1889468 used,    30064 free,   114448 buffers
> Kb Swap:  5927948 total,    93216 used,  5834732 free,   457912 cached
> 
>  PID USER      PR  NI  VIRT  RES  SHR S  %CPU %MEM    TIME+  COMMAND
> 31309 sm        39  19  295m 209m 5020 S 142,5 11,2  37:53.38 virtuoso-t

That's ridiculous.  No database manager worth its salt in this day and age
should get bogged down in its own housekeeping.  Maybe Nepomuk/Strigi
should switch to another DBMS, such as the one already used in Akonadi,
and save on overheads, downloads, installation, memory, disk space, etc.

Anyway, however good N/S may be (and I have yet to see that) I do not want
a search engine that would drain my laptop battery.  Nepomuk/Strigi is
supposed to be a "Pillar of KDE" but to me it looks as though it is standing
on wet clay.  I doubt if a nice search widget could lift it out of the doldrums,
Stefan, but I wish you the best of success with your initiative.

All the best, Ian W.



>> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe <<
[prev in list] [next in list] [prev in thread] [next in thread]