[CLOSED] Firebird-SQL

[rvk]

He could also just create fields for those KEY_NAMEs in SAMPLES itself
if for every record in SAMPLES there is only one record per KEY_NAME.

But that goes against the principle of normalizing the database 

[Zvoni]

I think he wrote early on, that he can‘t do that since the „column“-name (key) might change

[rvk]

I think he wrote early on, that he can‘t do that since the „column“-name (key) might change

Yes

Yes, this design is ugly. I've not implemented yet. Just thinking over. Problem is the "property" names may change, so that I cannot fix fields at design time.

But when that happens... ANY SQL with fixed values of KEY_NAME will become invalid too.

Yes, you can put those fields in a json blob, but you'll loose the ability to do a select on them (if you used the values).

And putting only the key_names there is kind of useless because a simple

Code: SQL  [Select][+][-]

1. SELECT DISTINCT KEY_NAME FROM SAMPLE_DATA

will retrieve the key names too after which you can build the SQL.

[egsuh]

Thanks a lot for all the advices.
There would be many projects, and the key_names will differ from project to project, but the same within each project.
I can manage it by creating or altering a view for each project. Record numbers would not be "very" large. Possibly there might be several hundred K projects, but in most cases hundreds or thousands, at most. Only have to write "saving" procedures. Many thanks again.