[REBOL] Re: ANN: RebDB Pseudo-Relational Database
From: atruter:labyrinth:au at: 28-Jan-2004 10:48
> DefiantPC has been good enough to add a RebDB forum at
Thanks, I'm using that now.
> I would like to use an AltME group for this.
REBOL Talk got in first, and it's a bit more accessible IMHO.
> Now it will be interesting to see performance comparison of your
> database vs. DyBase one
Not really. They have radically different design goals and
implementation's, and are optimised for different tasks and data volumes
(RebDB being optimised for simple searches in the 10,000 - 1,000,000
> Wouldn't work. I read that RebDB uses fixed length fields
> whereas I need varchar type fields.
is true if you need to store something like the email body
text *and* you want the DB to operate on it with statements such as
soundex, like, etc. If, however, it is the header data you need to operate
on and you merely need access to the underlying body text when specific
records are retrieved then storing file! references to individual text
files (perhaps compressed) has its own advantages: among them the fact
that incremental backups are a lot easier!
I'm not trying to convince anyone of anything in particular, it's just
that I've worked with enough DBA's to know that sometimes storing
*everything* in the database is not the best solution: the more successful
solutions I've come across play to the strengths of both the DB *and* the
underlying file system.
On a final note, thanks for all the positive feedback. We now return you
to your regular program schedule (please move any future RebDB discussion
to the RebDB forum at REBOL Talk).