Re: [Emacs-diffs] trunk r113958: * minibuffer.el (completion--sifn-requote): Bind `non-essential'.

[Stefan Monnier]

>>> * minibuffer.el (completion--sifn-requote): Bind `non-essential'.
>> This is wrong.  completion--sifn-requote is used in many contexts, some
>> of which are "essential" and some not.  IOW if you need to bind
>> non-essential, it shouldn't be here, but somewhere higher up the
>> call-chain.
> Honestly, I don't know too much about the completion code in
> minibuffer.el. The OP has reported the following backtrace (shortened by
> me):

My point is simply that the backtrace you show does not seem to perform
non-essential work, so binding non-essential is not right.  It is
computing the completion of /ssh:address@hidden:/ (not quite sure
why, since the backtrace doesn't go further up).

> Sadly, I cannot reproduce this scenario locally. And I have no idea, why
>       completion--sifn-requote(27 "/ssh:address@hidden:/")
> calls
>       substitute-in-file-name("/ssh:address@hidden")

Can't say offhand either why not.  Maybe we should simply demote errors
from substitute-in-file-name inside completion--sifn-requote.

At the same time I wonder why (substitute-in-file-name
"/ssh:address@hidden") should signal an error: it looks like an
incomplete filename, so we don't know yet whether it'll really be an
error or not.


        Stefan