[Mulgara-general] Is storing images practical?

[David Legg]

Hi Paul,

Thanks for responding (twice :-) ).

It's good to know that the underlying infrastructure should be able to 
handle large chunks of data.  As you might have guessed in my second 
email (the one that arrived first and thus gave me time to re-state my 
question) I'm more concerned with getting images and HTML content into 
or out of the triplestore rather than trying to compare or sort them in 
a query.  I'm glad you mentioned how slow a query involving large data 
might be.  The thought hadn't occurred to me but it is definitely 
something to watch out for.

I'm trying not to have to run a triplestore AND a content repository 
(e.g. JCR) or traditional database when building a web site.  I think a 
triplestore should be able to do the work of a content repository (JCR) 
in addition to a repository of semantic data but it all depends on 
whether the triplestore can efficiently process 'content-sized' chunks 
of data and not just little strings containing a person's name or their 
integer age for example.

> How are you accessing this data right now? Using the Connection API
> directly should allow you to avoid encoding and decoding, while HTTP
> will force it on you. Also, if you're using the Connection API with
> RMI (which you have to, unless the database is embedded) then it may
> not want to handle large blocks of data. We do stream results back in
> chunks, but literals don't get broken up.
>   

Mmm... that sounds ominous.  You do allow the entire triplestore to be 
loaded using a load command via the connection API for example: -

  connection.execute(*new* Load(dataFile, myData, *true*));


So I assumed this wouldn't be a problem.  I guess I'll have to try it ;-)

Regards,
David Legg

P.S.  Sorry for the tardy reply... lots of things to do this weekend 
including taking my son to see Avatar... well recommended in 3D!