Values from sequences (e.g. postgres' serial datatype) can (validly) vary and will often thwart the otherwise so attractively simple dump & diff approach.
And IMHO tobyink is right: without index, order is not garantueed, and I'd consider it unwise to rely on order in the database on the basis of the order of the source material.
Just a general remark: the OP doesn't give much detail, and doesn't mention what system he works with, but in case the system is PostgreSQL: it has a md5() function which can be handy for directly comparing data (and searching for differences via SQL).
In reply to Re^4: Testing database updates.
by erix
in thread Testing database updates.
by chrestomanci
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |