Today, Doug Alcorn <[EMAIL PROTECTED]> wrote:
> What would keep "us" (read "the people who develop bbdb") from
> implementing a more orthogonal schema in a flat file? I agree with
> the other two posts about not using either RDBM or something like
> Berkley db files. It is just too much hastle to setup for not much
> gain. However, I also think that something like the above proposed
> schema would be more flexible. You could basically segment the file
> into table sections with "Beginning of Table" and "End of Table"
> markers. This would deviate quite significantly from the current "one
> line is a complete record" format. Are there any real drawbacks to
> this?
Jup. Maintaining this db in anything else than in memory (say: on disk)
is The Horrible Thing That You Don't Want to Have to Do. (-:
Seriously though, an implementation in several files (one per table)
would be OK. It would require less data to be read at startup time (read
the person table at startup, load the rest -- esp. the notes table --
when it's needed), and reduce search times -- you need to grep through
one file only (though it's arguable if the file is not already loaded
into memory).
Lack of time restricts me from implementing this right now (and for the
next um. 7 months). The data model is there if you want to do something
with it (-:
regards,
--
Andreas Fuchs, <[EMAIL PROTECTED]>, <[EMAIL PROTECTED]>, antifuchs
_______________________________________________
bbdb-info mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/mailman/listinfo/bbdb-info
