Re: [PATCH] Fix `early-init-file' value when file is missing

[Eli Zaretskii <eliz@gnu.org>]

> From: Radon Rosborough <radon.neon@gmail.com>
> Date: Fri, 8 Feb 2019 09:34:08 -0800
> Cc: emacs-devel <emacs-devel@gnu.org>
> 
> > However, I would prefer to have this solved outside
> > load-user-init-file, if feasible. This function is too central to
> > Emacs to make non-trivial changes there for such a minor problem's
> > sake.
> 
> I am sorry, but I simply cannot see how this change is non-trivial.
> There is exactly one logic change that this patch makes: namely, that
> the extension of user-init-file is changed to .el, but only in the
> case of the early init-file not being found.

But to do that, you've modified the API of load-user-init-file,
changing the semantics of its second argument, which then caused the
code of the function to change to accommodate that.  This then
requires to go and check that these changes didn't affect the other
callers of that function in ways we don't want.

> I also cannot see how any other way of solving this bug could be as
> clear and straightforward as fixing it directly by updating the code
> that was broken.

All you want is to set a single variable to a specific value if the
file wasn't found, right?  How hard can it be to do that after
load-user-init-file returns and reports a failure?  Or do that inside
the function, but only if it processes early-init file specifically?

> Could you explain your reasoning?

I tried in the message you responded to, but you've elided all those
explanations.