Re: [HACKERS] how to deal with sparse/to-be populated tables

Alfred Perlstein wrote:
> > There are two things you can do...
> >
> > 1) remember if a record came out of the database in the first place
> > with a flag. This is what an object database would do.
>
> You mean implement an LRU cache outside the database, I've thought about
> this and could actually do it, the thing that bugs me about it is
> that i'm essentially trying to outsmart a 10+ year (guessing) old
> piece of software with something that I'd have to hack up in a
> matter of days.

Well, you only gave a small code snippet, I don't know how your app
works.

But often you retrieve tuples from the database and populate a C struct
or something...

struct Person {
char *firstname;
char *lastname;
};

What I'm saying is, if you are already doing something like this, then
just add one more boolean to say if it is a new or existing Person.
If you are not doing anything like this currently then it's not an
option.

Alternatively wait for my ODBMS features :-)

> This is what I was thinking, the problem then becomes that I'm
> not aware of way to determine the error with
> some degree of accuracy so that I don't mistake:
> insert error because of duplication
> with:
> insert error because of database connectivity (or other factors)
>
> Is it possible to do that? I guess I could parse the error responce
> from the backend, but maybe there's an easier/more-correct way?

Hmm. Doesn't PostgreSQL have a big list of error codes? I don't think
it does, I've never seen one. There should be a way to get error
codes without comparing strings. Should this be on the TODO?