Re: pgsql: Clarify coding of .exe patch

Tom Lane wrote:
> Bruce Momjian <pgman@candle.pha.pa.us> writes:
> > !     if (strlen(nodir_name) > sizeof(EXE) - 1 &&
> > !         pg_strcasecmp(nodir_name + strlen(nodir_name)-sizeof(EXE)-1, EXE) == 0)
>
> This is clearer than "4"?
>
> If there were any remote chance that Microsoft would invent some new,
> different-length spelling of the ".exe" suffix in the foreseeable
> future, I might agree that this was good code.  As is, it looks like
> unhelpful obscurantism, because it requires a very close reading of the
> details of the C spec to confirm that it is correct and not wrong.
>
> Code that looks wrong on its face is not an improvement over code that
> is readable but doesn't cover changes that won't ever happen.

The '4' to me just looks too arbitrary --- the EXE reference makes it
clear to me what it.  I don't like arbitrary constants in the code and
especially not repeated several places.

As I just posted I would like us to create a global define that is
sizeof()-1 so it would be clear, perhaps CC_STRLEN() as a define name.

--
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 359-1001
  +  If your life is a hard drive,     |  13 Roberts Road
  +  Christ can be your backup.        |  Newtown Square, Pennsylvania 19073