Robert Haas wrote:
> On Tue, Sep 21, 2010 at 7:05 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> > Robert Haas <robertmhaas@gmail.com> writes:
> >> I don't understand the argument that we need type input functions to
> >> be protected by a savepoint. ?That seems crazy to me. ?We're taking a
> >> huge performance penalty here to protect against something that seems
> >> insane to me in the first instance. ?Not to mention cutting ourselves
> >> off from really important features, like the ability to recover from
> >> errors during COPY. ?I don't understand why we can't just make some
> >> rules about what type input functions are allowed to do.
> >
> > There are many rules that you could possibly make for type input
> > functions. ?But "you cannot throw an error" is not one of them ---
> > or at least, not one that you can usefully expect to be followed
> > for anything more than trivial straightline code.
>
> OK. This is one of the things I don't understand. Why does throwing
> an error imply that we need to abort the current transaction? Why
> can't we just catch the longjmp() and trundle onwards? Obviously,
> that's unsafe if a pretty wide variety of cases, but if you're just
> scrutinizing the input string (even with a little bit of read-only
> database access) it's not obvious to me what can go wrong. (I assume
> there is something, but I don't know what it is...)
That would be interesting. You would need to flag that this was not a
longjump requiring cleanup, but the harder part would be getting back to
where the error occurred. I guess you could rerun the query. :-|
-- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB
http://enterprisedb.com
+ It's impossible for everything to be true. +