> Neil Padgett <npadgett@redhat.com> writes:
> >> I like the idea of putting the formatting stuff in libpq. It
> >> centralizes it, and allows the client to control the formatting too.
>
> > What exactly would you put in libpq?
>
> I think we could put in code that parses the multi-line error message
> format, and returns preparsed data in the form of a list of field names
> and field values. I don't see that libpq can do anything useful with
> producing a syntax-error pointer, since it doesn't have access to the
> original user query string, only to the same string that's sent to the
> backend; so as far as it can know, the error index that the backend
> returns is gospel. Any reverse-mapping from that to a user-query index
> has got to be in the client app, AFAICS.
Seems we could easily have the backend return a string with three tokens
before/after the syntax error, and that would be enough. It would save
the client from doing all this housekeeping just for syntax errors.
--
Bruce Momjian | http://candle.pha.pa.us
pgman@candle.pha.pa.us | (610) 853-3000
+ If your life is a hard drive, | 830 Blythe Avenue
+ Christ can be your backup. | Drexel Hill, Pennsylvania 19026