You are not logged in.
Pages: 1
May I humbly suggest... that you modify your tables for storage to something ( like ) the following:
int ( id )
foreignkey ( self id )
int ( block id )
blob ( data )
Where the first col is a generated unique id... the foreign key is pointing back to an id of an object of the same type... the block id is to allowed proper ordering of all blocks where id == self id == ( some value ) ... more or less just a simplistic inode-ish type system...
The rational is that by limiting the data blob segment to a size that you know will work on ANY ( or a large number of ) db / driver combo impls... then it - in the end - makes it much simpler for any users / maintainers of the system... though it does add ( slightly ) more load on the db...
Of course...
I'm more of a defensive programmer than most... but there is a reason... I've just had to suffer thru too many times where product 'A' required a specific dependency version... and product 'B' required the same dependency but a different version ( or range of versions )... AND the thing was I wanted to use both products A and B but couldn't... so as a programmer I want users of my code to avoid the same sort of frustration if possible...
Michael
"The only difference between me and a madman is that I'm not mad."
Salvador Dali (1904-1989)
Offline
Pages: 1