Re: [Babeldoc-devel] Database Tables

[<ek...@ba...>]

Sounds interesting and flexible but perhaps a little complicated.  I
regret that I cannot immediately provide a more "elegant" solution, but
I can tell you what was discussed at my database design meeting
yesterday with our data architect ... please bear one thing in mind
before reading on ... I'm tasked with delivering a very challenging
system in a very short timeframe ... that being said, we have initially
decided to add a Sybase "text" field (unlimited character length field)
to the CONFIG table, named "cfg_script" that is nullable.  For rows in
this table in which you would normally reference a file descriptor or
URL in the cfg_value field, we will embed that script into "cfg_script"
and read it in using SQLConfig.getConfig().  I will then have to figure
out a way to feed this script (contained in a String) to the appropriate
process (Xalan, etc) ... but I haven't even thought that far yet.

Yes, your solution is MUCH more elegant than what I have detailed above,
but I've got management coming down on me like a ton of bricks to get
this project designed and developed.  I would much rather implement a
solution that keeps me in-line with Babeldoc's progress.

Thanks again for all your assistance and interest.

Erik

<<
Erik,  Dejan:

I think I have found a relatively simple solution:

1.  We implement a new URLStreamHandler so that we can have a new
protocol
(lets say, babelsql:) that can access the database just like a
filesystem.
2.  We implement the virtual filesystem using three tables - a directory
table
that knows about about its parent directory and maintains an id.  A data
table that knows about its directory table id and its own id and an
"inode"
table that tracks common stuff for the two tables: permissions, last
write,
access time, modification etc.
3.  Then we simply manage the data in the tables just like a regular
filesystem and we access them in the same way.

thoughts?

regards,
Bruce.
>>