Re: Moving other hex functions to /common

On Thu, Dec 31, 2020 at 03:10:29PM +0900, Michael Paquier wrote:
> On Wed, Dec 30, 2020 at 08:22:07PM -0500, Bruce Momjian wrote:
> > So, I am learning this cfbot thing.  Seems I need -M100% to disable
> > rename detection for diffs to work with cfbot --- makes sense.
> 
> A good way to make sure that a patch format is correct for the CF bot
> would be to use "git format-patch -1" to generate a patch from a
> single commit.

Thanks.  I had to learn how to squash my commits into a new branch and
then generate a format-patch on that:

    https://bugsdb.com/_en/debug/8b648ec395b86be32efa9629cb006d74

I wanted to see how the cfbot liked my original patch with the renames,
and it didn't, so now I know I have to use this method for the
commitfest.  Patch attached.

> > New patch attached.
> 
> I think that this patch would have more value if we remove completely
> the hex routines from ECPG and have ECPG use what's moved in
> src/common/, meaning the following changes:
> - Remove the exit(), pg_log_fatal() and ereport() calls from
> src/common/hex.c, replace the error code paths with -1, and return a
> signed result.
> - The ECPG copies make no use of ecpg_raise(), so things map easily.
> - This means keeping small wrappers in encode.c able to generate those
> ereport(FATAL) in the backend, but that's just necessary for the
> decode routine that's the only thing using get_hex().
> - Let's prefix the routines in src/common/ with "pg_", to be
> consistent with base64.
> - It would be good to document the top each routine in hex.c (see
> base64.c for a similar reference).

Let me get my patch building on the cfbot and then I will address each
of these.  I am trying to do one stage at a time since I am still
learning the process.  Thanks.

-- 
  Bruce Momjian  <bruce@momjian.us>        https://momjian.us
  EnterpriseDB                             https://enterprisedb.com

  The usefulness of a cup is in its emptiness, Bruce Lee