Re: [Emacs-diffs] master 1fcc552: lisp/gnus/registry.el (registry-prune): Allow registry to reach full size before pruning

Re: [Emacs-diffs] master 1fcc552: lisp/gnus/registry.el (registry-prune): Allow registry to reach full size before pruning

[Stefan Monnier]

> +    (if (> size (oref db :max-size))

Why do people use the :initarg rather than the slot name?

   (oref db max-size)

works as well (actually better: it's implemented more efficiently).


        Stefan

Re: [Emacs-diffs] master 1fcc552: lisp/gnus/registry.el (registry-prune): Allow registry to reach full size before pruning

[Eric Abrahamsen]

Stefan Monnier <monnier@IRO.UMontreal.CA> writes:

>> +    (if (> size (oref db :max-size))
>
> Why do people use the :initarg rather than the slot name?
>
>    (oref db max-size)
>
> works as well (actually better: it's implemented more efficiently).

Because we didn't know any better! For some value of "us"...

I saw your recent patch to the registry files, but you only changed
*some* of the calls to oref/oset to use slot names. That was confusing.
I can change the rest of them over, if that's desirable.

Re: [Emacs-diffs] master 1fcc552: lisp/gnus/registry.el (registry-prune): Allow registry to reach full size before pruning

[Stefan Monnier]

> I saw your recent patch to the registry files, but you only changed
> *some* of the calls to oref/oset to use slot names. That was confusing.

I have a local change which signals an error when such an `oref' is
executed (but not when it's compiled).  So I only bump into some
of them.

> I can change the rest of them over, if that's desirable.

Yes, I hope you (or other people) can take it from there.


        Stefan

Re: [Emacs-diffs] master 1fcc552: lisp/gnus/registry.el (registry-prune): Allow registry to reach full size before pruning

[Eric Abrahamsen]

Stefan Monnier <monnier@IRO.UMontreal.CA> writes:

>> I saw your recent patch to the registry files, but you only changed
>> *some* of the calls to oref/oset to use slot names. That was confusing.
>
> I have a local change which signals an error when such an `oref' is
> executed (but not when it's compiled).  So I only bump into some
> of them.
>
>> I can change the rest of them over, if that's desirable.
>
> Yes, I hope you (or other people) can take it from there.

Done!

Re: [Emacs-diffs] master 1fcc552: lisp/gnus/registry.el (registry-prune): Allow registry to reach full size before pruning

[Ted Zlatanov]

On Fri, 20 Mar 2015 10:35:59 -0400 Stefan Monnier <monnier@IRO.UMontreal.CA> wrote: 

>> +    (if (> size (oref db :max-size))
SM> Why do people use the :initarg rather than the slot name?

SM>    (oref db max-size)

SM> works as well (actually better: it's implemented more efficiently).

I used it because I found it in a manual.  But I can't recall which
manual and when, it's been a while.  AND IT WORKED DAMMIT! :)

If initargs are to be avoided, maybe a compilation warning there would
be appropriate?[1]

Ted

[1] this is where you tell me one has been in place since 1992 :)

Re: [Emacs-diffs] master 1fcc552: lisp/gnus/registry.el (registry-prune): Allow registry to reach full size before pruning

[Stefan Monnier]

> If initargs are to be avoided, maybe a compilation warning there would
> be appropriate?[1]

Agreed.  I haven't implemented one yet, but it's crossed my mind, yes.
The main problem is that in (oref a b), the compiler has no idea what is
the possible class(es) of `a', so it can't look up that class to see if
`b' is an initarg.

What I'm thinking of doing is keeping track of all known slot names and
all known initarg names, and if `b' is an initarg name but not a slot
name, then emit the warning.  This should work OK in practice since
initargs tend to all be keywords, where slot names tend to all be
non-keyword symbols.


        Stefan

Re: [Emacs-diffs] master 1fcc552: lisp/gnus/registry.el (registry-prune): Allow registry to reach full size before pruning

[Ted Zlatanov]

On Sat, 18 Apr 2015 10:20:30 -0400 Stefan Monnier <monnier@IRO.UMontreal.CA> wrote: 

>> If initargs are to be avoided, maybe a compilation warning there would
>> be appropriate?[1]

SM> Agreed.  I haven't implemented one yet, but it's crossed my mind, yes.
SM> The main problem is that in (oref a b), the compiler has no idea what is
SM> the possible class(es) of `a', so it can't look up that class to see if
SM> `b' is an initarg.

SM> What I'm thinking of doing is keeping track of all known slot names and
SM> all known initarg names, and if `b' is an initarg name but not a slot
SM> name, then emit the warning.  This should work OK in practice since
SM> initargs tend to all be keywords, where slot names tend to all be
SM> non-keyword symbols.

That would work for me as a user, but I fear it will be pretty hard to
implement.

Ted