The Interesting Installation Integer Problem

[Gervase Markham]

Max Kanat-Alexander wrote:
> On Mon, 2005-03-07 at 20:28 +0000, Gervase Markham wrote:
> 
>>I may be mistaken, but isn't this what the SQL standard does - i.e. 
>>define a set of types with names that all databases understand?
> 
> 	I wish that all databases understood them. And that we could then get
> those exact field names back with DBI's type_info. Unfortunately,
> neither of those things are true.

I would have thought that the latter could be made true without too much 
effort by using our wrapper around DBI to look up a name in a table and 
convert it.

And it seems odd that databases claim to support SQL and yet don't 
understand the defined data types...

> 	The two won't get out-of-sync -- the "ChangeField" subs will only
> change the stored schema after a successful ALTER TABLE call. Since
> those things always happen atomically (we never call ALTER TABLE outside
> of a "ChangeField" call), there's almost no risk of them getting out-of-
> sync.

Our code may be that scrupulous... but other people's might not. OK, so 
we could claim that people hacking the database are on their own and 
deserve what they get, but we haven't claimed that thusfar.

Gerv