Re: WIP: a way forward on bootstrap data

On 3/22/18, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Looking at this again, I think a big chunk of the readability problem here
> is just from the fact that we have long, similar-looking lines tightly
> packed.  If it were reformatted to have comment lines and whitespace
> between, it might not look nearly as bad.
>
...
> Anyway, for the moment I'd stick with BKI_LOOKUP rather than undoing
> that work.  I think it's a more transparent way of saying what we
> want than the magic-OID-typedefs approach.  The formatting issue is
> just a mild annoyance, and it's not really BKI_LOOKUP's fault anyway.

Okay, I'll do it with comments and whitespace.

> While I'm thinking of it --- I noticed one or two places where you
> had "BKI_DEFAULT(\0)".  That coding scares me a bit --- gcc seems to
> tolerate it, but other C compilers might feel that \0 is not a valid
> preprocessing token, or it might confuse some editors' syntax highlight
> rules.  I'd rather write cases like this as "BKI_DEFAULT('\0')".  IOW,
> the argument should be a valid C identifier, number, or string literal.

Hmm, I only see this octal in pg_type.h

char        typdelim BKI_DEFAULT(\054);

Which I hope is fine. Were you thinking of this comment in
pg_attribute.h? We use the double-quoted empty string for postgres.bki
and change it to  '\0' for schemapg.h.

/* One of the ATTRIBUTE_IDENTITY_* constants below, or '\0' */
char        attidentity BKI_DEFAULT("");


-John Naylor