On Tue, May 24, 2022 at 11:41:49AM -0700, Nathan Bossart wrote:
> I would actually ERROR on this so that we aren't relying on
> --enable-cassert builds to catch it. That being said, if there's no strong
> reason to enforce that a short description be provided, then why not adjust
> ShowAllGUCConfig() to set that column to NULL when short_desc is missing?
Well, issuing an ERROR on the stable branches would be troublesome for
extension developers when reloading after a minor update if they did
not set their short_desc in a custom GUC. So, while I'd like to
encourage the use of short_desc, using your suggestion to make the
code more flexible with NULL is fine by me. GetConfigOptionByNum()
does that for long_desc by the way, meaning that we also have a
problem there on a build with --enable-nls for short_desc, no?
--
Michael