Hi, On Thu, Dec 04, 2025 at 10:34:34AM +0100, Daniel Gustafsson wrote: > > On 2 Dec 2025, at 15:28, Bertrand Drouvot <[email protected]> > > wrote: > > > That said I'm still skeptical that we need to provide a strong > > justification (as the one above) to remove an unused parameter. > > If it breaks an existing published API thus causing extensions to fail to > compile then IMHO that's a pretty strong argument against removing a parameter > even if it's unused,
Yeah, that's why I did focus on static functions only. > likewise if the change can be expected to cause backpatching conflicts for > the coming five years. Fair point about backpatching. That said, I did find examples of commits removing unused parameters (see above). I'm trying to understand when this kind of cleanup is considered acceptable vs. when the backpatching cost outweighs the benefit. Any guidance would be helpful. > For static functions at > least it seems that compilers are fairly happy to remove the parameter in > greater than -O0 levels (though I know that won't move the needle on one of > your main drivers being readability). Yeah, my motivation isn't execution efficiency. Regards, -- Bertrand Drouvot PostgreSQL Contributors Team RDS Open Source Databases Amazon Web Services: https://aws.amazon.com
