Обсуждение: Patch for some bugs/annoyances

Here's a patch fixing a bug in throwing a PSQLException which causes
helpful information to not be displayed. The code is clearly trying to
call the PSQLException(String, PSQLState, Object[]) constructor, but
instead gets the PSQLException(String, Object) constructor because it
forgets the PSQLState.

This first bug may have been causing the second; I can't remember
anymore. Anyway, sometimes this.state in PSQLException is null causing
getSQLState() to throw a NullPointerException. I've included a check for
this and return the string "null" when the state is null.

Finally, I've changed PSQLState to implement Serializable so that I can
throw SQLExceptions to remote callers.

On Thu, 11 Dec 2003, Patrick Higgins wrote:

> This first bug may have been causing the second; I can't remember
> anymore. Anyway, sometimes this.state in PSQLException is null causing
> getSQLState() to throw a NullPointerException. I've included a check for
> this and return the string "null" when the state is null.
>

I don't think returning "null" is correct.  I think
PSQLState.UNKNOWN_STATE should be used as the state when it is null.

Kris Jurka

On Thu, 2003-12-11 at 14:32, Kris Jurka wrote:
> On Thu, 11 Dec 2003, Patrick Higgins wrote:
>
> > This first bug may have been causing the second; I can't remember
> > anymore. Anyway, sometimes this.state in PSQLException is null causing
> > getSQLState() to throw a NullPointerException. I've included a check for
> > this and return the string "null" when the state is null.
> >
>
> I don't think returning "null" is correct.  I think
> PSQLState.UNKNOWN_STATE should be used as the state when it is null.

That sounds like a much better idea. I only used "null" because I didn't
know better.

On Thu, 11 Dec 2003, Patrick Higgins wrote:

> On Thu, 2003-12-11 at 14:32, Kris Jurka wrote:
> > On Thu, 11 Dec 2003, Patrick Higgins wrote:
> >
> > > This first bug may have been causing the second; I can't remember
> > > anymore. Anyway, sometimes this.state in PSQLException is null causing
> > > getSQLState() to throw a NullPointerException. I've included a check for
> > > this and return the string "null" when the state is null.
> > >
> >
> > I don't think returning "null" is correct.  I think
> > PSQLState.UNKNOWN_STATE should be used as the state when it is null.
>
> That sounds like a much better idea. I only used "null" because I didn't
> know better.
>

Here's a patch to that effect.  Also the Serializable part has already
been addressed.

Kris Jurka