[DGD]serialisation/deserialisation of mixed data.

[Felix A. Croes]

balduin at uni-paderborn.de wrote:

>[Zellski wrote:]
> > If you are used to writing LPC, there's a fair chance that it'd be quicker
> > to write a SQL*Net implementation than to write these kfuns in C. I don't
> > know how proprietary the protocol is, but SQL and the DB functionality
> > shouldn't be too hard to do?
>
> Well it's not the problem to retrieve the data. I wrote some kfun wrapper
> functions to access the mysql - C API with lpc. The problem why i want to
> store dgd-objects in the database and not in the state-dump of dgd is the
> limited number of objects dgd can handle, 64k seems to be a lot, but from 
> Xyllomer i know that at reboot time (after 24h) we usually have about 20k
> objects.  If I use a state-dump reboot won't help a lot to decrease this 
> number and i'd be limited to about 10k Documents or even less, if I assume
> that modern html pages consist of a lot of subdocuments e.g. pictures.
> This is a lot for a single class, but even the document server of our 
> workgroup at my university has about 10k documents. Not talking about
> links, personal annotations and such stuff of the students. This forces
> me to kick objects out of memory if they are not used for a period of
> time which imposes a part of the problem i described, and makes the
> state-dump useless for me.

DGD can handle more than 64K objects if you make the following change
in dgd/src/config.h: change the typedef for uindex into unsigned long,
and the #define for UINDEX_MAX into ULONG_MAX.

A side effect of this change is that the server will use more memory
overall.

Regards,
Dworkin

List config page:  http://list.imaginary.com/mailman/listinfo/dgd