Quoting Igor Tandetnik <itandet...@mvps.org>:
stahl...@dbs.uni-hannover.de wrote:
SQLite's behavior makes sense, because *every* column type may be left out.
However, I think that in the case of FK-definitions (like the one in 'tab2')
assigning the default type is not the right thing to do.
Why should one clause in the syntax behave differently depending on
the presence or absence of another, unrelated clause? Sounds like an
arbitrary special case to me.
I think that's the main point of our disagreement.
In my opinion the column type definition is very well related to the
foreign-key definition:
Barring more specific clauses, a column defined as a foreign key
should exactly
be like the primary key it references -- including data type, collation rules
and any other potential modifiers.
This, as I see now, collides with SQLite's view that leaving out a
column type,
actually means "BLOB".
As no one here seems to agree with my view, I will no further pursue this.
(Also I'm quite alright with specifying the column type explicitly
every time.)
My initial 'bug'-report now becomes a well-meant suggestion at best. :-)
Thank you very much for your insight!
Christian
_______________________________________________
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
