Re: pg_get_functiondef forgets about most GUC_LIST_INPUT GUCs

Hello,

At Wed, 14 Mar 2018 17:46:21 +0900, Michael Paquier <michael(at)paquier(dot)xyz> wrote in <20180314084621(dot)GA617(at)paquier(dot)xyz>
> On Wed, Mar 14, 2018 at 05:30:59PM +0900, Kyotaro HORIGUCHI wrote:
> > Doesn't it make sense if we provide a buildtime-script that
> > collects the function names and builds a .h file containing a
> > function using the list?
> >
> > The attached perl script is a rush work of such script, which
> > works at the top of the source tree. It just prints the function
> > definition, does not generate a .h file.
> >
> > I haven't confirmed anything about it but I had the following
> > output from the current master.
>
> I quite like that idea actually. Please note that pl_handler.c declares
> two more of them, so we may want a smarter parsing which takes into
> account declarations of DefineCustom*Variable.

Only DefineCustomStringVariable can accept a list argument even
if flagged GUC_LIST_INPUT. (I'm not sure this should be
explicitly rejected.) I'm not sure this is smart enough but it
works. It would fail if description for variables contain the
same character sequence to the lexical structure around but it is
rare to happen.

The attached patch is the result. It adds a script to generate
include/common/check_listvars.h. It won't be updated even if
related files are modifed (since we cannot know it before
scanning the whole source.) but "make clean" removes it. Regtests
is a copy of Michael's v1 patch.

regards,

--
Kyotaro Horiguchi
NTT Open Source Software Center