Path: ...!news.roellig-ltd.de!news.mb-net.net!open-news-network.org!eternal-september.org!feeder3.eternal-september.org!news.eternal-september.org!.POSTED!not-for-mail
From: Lawrence D'Oliveiro <ldo@nz.invalid>
Newsgroups: alt.folklore.computers,comp.os.linux.misc
Subject: Re: Multiple Inheritance [was Re: The joy of FORTRAN]
Date: Tue, 1 Oct 2024 21:47:52 -0000 (UTC)
Organization: A noiseless patient Spider
Lines: 61
Message-ID: <vdhqm7$2t1fi$3@dont-email.me>
References: <pan$96411$d204da43$cc34bb91$1fe98651@linux.rocks>
	<5mqdnZuGq4lgwm_7nZ2dnZfqnPSdnZ2d@earthlink.com>
	<vcub5c$36h63$1@dont-email.me> <1r0e6u9.1tubjrt1kapeluN%snipeco.2@gmail.com>
	<vcuib9$37rge$5@dont-email.me> <vcvuhh$3hroa$2@dont-email.me>
	<llhieuF8ej2U2@mid.individual.net> <20240925083451.00003205@gmail.com>
	<Pascal-20240925164718@ram.dialup.fu-berlin.de>
	<mdd4j63pmo1.fsf_-_@panix5.panix.com>
	<oJ-cnQSrLZDYdGX7nZ2dnZfqnPWdnZ2d@earthlink.com>
	<vdatb6$1l4ch$8@dont-email.me> <vdauah$1lq1u$1@dont-email.me>
	<lltt6uF4fseU4@mid.individual.net> <vdcn1q$1tmdr$5@dont-email.me>
	<vddqoe$264fi$2@dont-email.me> <vdf4le$2cn51$6@dont-email.me>
	<vdh2l1$2p7e2$1@dont-email.me>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Injection-Date: Tue, 01 Oct 2024 23:47:52 +0200 (CEST)
Injection-Info: dont-email.me; posting-host="45d68bef63035d0f49773effd7a1ab82";
	logging-data="3048946"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX1/lqIpuS0SelI713H+3K2Gw"
User-Agent: Pan/0.160 (Toresk; )
Cancel-Lock: sha1:oSheFm6855XCxSbrXOOr/K4T1cE=
Bytes: 4336

On Tue, 1 Oct 2024 15:57:36 +0100, Pancho wrote:

> On 9/30/24 22:19, Lawrence D'Oliveiro wrote:
>
>> But it makes the difference between multiple inheritance that bewilders
>> and confuses people, and multiple inheritance which makes sense. It’s
>> what lets Python have multiple inheritance that makes sense.
> 
> There is a difference between a consistent implementation of multiple
> inheritance and something that people actually understand when coding,
> or maintaining code.
> 
> In essence just because you can do something clever, doesn't mean you
> should.

A clever idea worth using is one that saves programmer work and 
implementation overhead.

For example, consider enumerated types. Java and C++ have to define these 
as part of the core language. But Python does not: enums are provided by a 
standard library module, written in pure Python -- there is nothing it 
does that you cannot do in your own Python code.

This enum implementation is built on two important ideas from the core 
language -- multiple inheritance, and metaclasses -- in order to do its 
thing. As a result, it can do everything that Java enums can do, for 
example, and more besides. And the result is very easy to use: you don’t 
need to understand all the intricacies of the mechanisms concerned to 
define simple enums.

(Want examples? Just ask.)

>> Think about why languages like Java and PHP avoided multiple
>> inheritance, and substituted those lame “interfaces” instead: it was to
>> avoid this bewilderment and confusion that is known to plague C++,
>> because the right solution wasn’t known at the time.
>> 
> I quite like interfaces, polymorphism through interfaces.

“Interfaces” are just “abstract base classes lite”. They are a way to 
provide a very limited form of multiple inheritance, in the hope that this 
would somehow avoid the bewilderment that comes from multiple inheritance 
as done in C++.

Do they succeed? You yourself said “Single inheritance bewilders and 
confuses people”, so I don’t think they did.

> I had a look, it would take me a while to work through the algorithm,
> establish how the ideas sit in the more common language of DAGs and
> graph theory.

The gist of it is, the linearization algorithm from CLOS was refined a bit 
for Dylan. And C3 is in turn a refinement of the linearization used in 
Dylan.

> If there is a significant difference that requires new
> terms such as monotonic.

“Monotonic” was not a new term in OO at the time of publication of that 
paper. In fact, the concept exposed some of the deficiencies that C3 
sought to address.