Goops: Setting Immutible Classes?

Goops: Setting Immutible Classes?

[Christopher Howard]

[-- Attachment #1: Type: text/plain, Size: 514 bytes --]

Hi, is the a way I could use goops such that I the classes are non-
mutating, but have setters that simply return a new instance of the
class (instead of modifying the original)?

I suppose I could simply use define-method to create setter methods
which create a new instance of the object; but that would get rather
complicated, wouldn't it, if the object had superclasses? So I need
some kind of shallow copy mechanism for the whole object, or a chained
copy mechanism...?

-- 
https://www.debian.org/

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: Goops: Setting Immutible Classes?

[Mark H Weaver]

Christopher Howard <christopher.howard@qlfiles.net> writes:

> Hi, is the a way I could use goops such that I the classes are non-
> mutating, but have setters that simply return a new instance of the
> class (instead of modifying the original)?

We provide such a mechanism for SRFI-9 Records.  Search for "Functional
Setters" in section 6.6.16 (SRFI-9 Records) of the Guile manual.

> I suppose I could simply use define-method to create setter methods
> which create a new instance of the object; but that would get rather
> complicated, wouldn't it, if the object had superclasses? So I need
> some kind of shallow copy mechanism for the whole object, or a chained
> copy mechanism...?

For GOOPS classes, take a look at the built-in 'shallow-clone' generic
function.  It is described in section 8.10 (GOOPS Object Miscellany).

      Mark

Re: Goops: Setting Immutible Classes?

[Thompson, David]

On Fri, Oct 6, 2017 at 2:00 AM, Christopher Howard
<christopher.howard@qlfiles.net> wrote:
> Hi, is the a way I could use goops such that I the classes are non-
> mutating, but have setters that simply return a new instance of the
> class (instead of modifying the original)?

I'm not too familiar with the meta-object protocol, but this sounds
like a case where you could make a new metaclass that supports
immutable slot setters.  See Section 8.11 in the manual.

- Dave

Re: Goops: Setting Immutible Classes?

[Christopher Allan Webber]

Mark H Weaver writes:

> Christopher Howard <christopher.howard@qlfiles.net> writes:
>
>> Hi, is the a way I could use goops such that I the classes are non-
>> mutating, but have setters that simply return a new instance of the
>> class (instead of modifying the original)?
>
> We provide such a mechanism for SRFI-9 Records.  Search for "Functional
> Setters" in section 6.6.16 (SRFI-9 Records) of the Guile manual.
>
>> I suppose I could simply use define-method to create setter methods
>> which create a new instance of the object; but that would get rather
>> complicated, wouldn't it, if the object had superclasses? So I need
>> some kind of shallow copy mechanism for the whole object, or a chained
>> copy mechanism...?
>
> For GOOPS classes, take a look at the built-in 'shallow-clone' generic
> function.  It is described in section 8.10 (GOOPS Object Miscellany).
>
>       Mark

See also
https://lists.gnu.org/archive/html/guile-user/2017-01/msg00030.html
which includes both a slot-fset approach and a clone macro.  Personally
I've come to like the clone macro more over time.