Comment 2 for bug 1066888

Revision history for this message
Dan Scott (denials) wrote :

Hi Bill:

Finally reviewing this - I like where you're taking things! A couple of thoughts with an eye towards using these in tests sooner rather than later:

1) Rather than using /* ... */ comments for the functions, why not use COMMENT ON FUNCTION - that way we can see the comments down the road right in the database.

2) It would be nice to mix in some non-ASCII user data into the names and addresses to ensure searching & displaying work as expected.

3) At least some of the users should have NULL dates of birth (many sites don't keep DOBs for their patrons) and rather than empty strings for suffixes we should probably use NULLs in most cases. (Having a few empty strings will add to the realism though).

4) Some (or perhaps most?) of the usrnames should be non-numeric, so that we can exercise code that determines barcodes vs. usernames. Perhaps we can reuse the passwords for this purpose?

5) This is my fault to begin with, but we should really use Dewey and LC call numbers so that we can properly test sorting and call number browsing.

I'll be happy to help out with these, following the rule of he who raises concerns gets to work on them... To reiterate, I really like what you're doing with this. Having the circ transactions in the mix really raises the bar!