Re: [PATCH] pgcrypto: Test for NULL before dereferencing pointer

Marti Raudsepp <marti(at)juffo(dot)org> writes:
> There's a file descriptor leak in psql/command.c function
> process_file() -- on errors it just returns without closing the file.
> But since it's quitting anyway, there's no practical impact. Should I
> submit a patch for this as well?

Might as well. It's the kind of thing that could turn into a real
bug given some rearrangement of the code.

> Then there's the 'badzero' check that finds a dozen cases where
> pointers are compared to a literal 0, not a NULL. This is a only a
> coding style check, as far as I can tell, so I thought it's not worth
> it.

I'd be in favor of fixing those too. I have no particular problem with
either "if (ptr)" or "if (ptr != NULL)", but I think that "if (ptr != 0)"
gives the reader entirely the wrong impression about the datatype of ptr.
Just because C fails to distinguish doesn't make it good style.

regards, tom lane