Re: define bool in pgtypeslib_extern.h

On Fri, Oct 25, 2019 at 9:55 PM Tom Lane <tgl@sss.pgh.pa.us> wrote:
>
> I wrote:
> > Amit Kapila <amit.kapila16@gmail.com> writes:
> >> As suggested by Andrew Gierth [1], I think we can remove the define in
> >> pgtypeslib_extern.h as it doesn't seem to be exposed.
>
> > Yeah, it's not good that that results in a header ordering dependency,
> > and it doesn't seem like a good idea for pgtypeslib_extern.h to be
> > messing with the issue at all.
> > If you like, I can experiment with that locally on prairiedog's host
> > before we make the buildfarm jump through hoops.
>
> I checked that that works and fixes the immediate problem, so I pushed
> it.
>

Thank you.

>  However, we're not out of the woods, because lookee here in
> ecpglib.h:
>
> #ifndef __cplusplus
> #ifndef bool
> #define bool char
> #endif                          /* ndef bool */
>
> #ifndef true
> #define true    ((bool) 1)
> #endif                          /* ndef true */
> #ifndef false
> #define false   ((bool) 0)
> #endif                          /* ndef false */
> #endif                          /* not C++ */
>
> #ifndef TRUE
> #define TRUE    1
> #endif                          /* TRUE */
>
> #ifndef FALSE
> #define FALSE   0
> #endif                          /* FALSE */
>
> This stuff *is* exposed to client programs, so it's not clear how
> painless it'd be to monkey around with it.  And it is used, further
> down in the same file, so we can't fix it just by deleting it.
> Nor can we import c.h to get the "real" definition from that.
>
> I'm more than slightly surprised that we haven't already seen
> problems due to this conflicting with d26a810eb.
>

I think it is because it never gets any imports from c.h.  It instead
uses postgres_ext.h.  If we want to fix this, the simplest thing that
comes to mind is to change the definition of bool in ecpglib.h and
probes.d to match with c.h.  These files contain exposed interfaces,
so the change can impact clients, but not sure what else we can do
here.  I have also tried to think about moving bool definition to
postgres_ext.h, but I think that won't be straightforward.   OTOH, if
you think that might be worth investigating, I can spend some more
time to see if we can do that way.

-- 
With Regards,
Amit Kapila.
EnterpriseDB: http://www.enterprisedb.com