|
Pages: [1]
|
 |
|
 Author
|
Topic: A general question on technique ... (Read 2153 times)
|
|
jfdutcher
|
I am converting a fixed record, fixed file size app and have started out by building a large nested block much like the db described in "Rebol, The Offical Guide". RebDB sounds much more appealing ....but while I have a 'maximum' of (300) rows...they would either have 120 odd columns or columns of 'blocks' which would reduce the # of rows considerably. Using your 'row-id' I believe I could add - delete indefinitely without a need to sort the table...I have a key field value I have to use for each. In the old system I pick the records and keep their physical record # with them , sort them, and then get the rest of the data using record # (or in RebDB, row-id). Over half the file is spaces - it's repeating data that either is, or is not, present.
I wonder if more than (1) RebDB table can be open at once..even though everything has been in one file in the old system I realize the
DB architect types would want to split the repeating values into
'related' tables....but if I can store blocks, even 'empty' blocks as a
column...maybe just as well. The current file occupys 789 kb on a
Windows system.
|
|
|
|
|
Logged
|
|
|
|
|
Ashley
|
RebDB supports blocks (and lists / hashs) as column values, and note that columns are not "type cast" so you can quite happily store different REBOL types within the same column.
RebDB allows you to have multiple tables open at the same time, but currently doesn't support JOIN type operations (allowing a single statement to access multiple tables at the same time).
|
|
|
|
|
Logged
|
|
|
|
|
|
Pages: [1]
|
|
|
 |
