permanent-local

permanent-local

[David Reitter]

Is there a reason why we don't have

	(put 'buffer-save-without-query 'permanent-local t)

?

Shouldn't such variables be permanent-local unless set by a major mode?

I don't quite understand why this sort of behavior is a function of  
the symbol (globally!) rather than of the value that is set: some  
major modes may set something like `buffer-save-without-query', but it  
may also be set buffer-locally by a minor mode.

Re: permanent-local

[Stefan Monnier]

> Is there a reason why we don't have

> 	(put 'buffer-save-without-query 'permanent-local t)

> ?

> Shouldn't such variables be permanent-local unless set by a major mode?

> I don't quite understand why this sort of behavior is a function of the
> symbol (globally!) rather than of the value that is set: some major modes
> may set something like `buffer-save-without-query', but it may also be set
> buffer-locally by a minor mode.

Yes, the `permanent-local' property is problematic every once in
a while.  Sometimes it's nil when you'd want it to non-nil and sometimes
it's the opposite.  Sadly it's a global setting and it's not clear how
to make the choice finer grained.

Maybe we could let `permanent-local' be not just a boolean but also
something else that would describe which *values* are permanent local
and which aren't.  So we could set buffer-save-without-query to either
t (not permanent-local) or `permanent' (behaves like t, but is
permanent-local).

In the case of buffer-save-without-query, I haven't seen enough uses to
know what we should do: the only use I see (in type-break.el) shouldn't
care either way.


        Stefan