Re: [Factor-talk] Abstraction like Clojure's protocols?

[Joe Groff]

On Thu, Nov 17, 2011 at 12:44 PM, missingfaktor  wrote:

> Factor's generic words are similar to Clojure's multimethods. However
> there are times when protocols (in Clojure speak) are more appropriate
> abstraction. Does Factor support a construct similar to protocols?
>

There is a protocol construct used by the delegation mechanism, which
groups together related generics:

GENERIC: succ ( n -- n' )
GENERIC: pred ( n -- n' )

PROTOCOL: enumerable succ pred ;

TUPLE: derived-enumerable from ;

CONSULT: derived-enumerable enumerable
from>> ;

It's not used as extensively in the Factor library as it perhaps should be.
In particular, protocols don't function as types.

-Joe
--
All the data continuously generated in your IT infrastructure 
contains a definitive record of customers, application performance, 
security threats, fraudulent activity, and more. Splunk takes this 
data and makes sense of it. IT sense. And common sense.
http://p.sf.net/sfu/splunk-novd2d___
Factor-talk mailing list
Factor-talk@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/factor-talk

[Factor-talk] Abstraction like Clojure's protocols?

[missingfaktor]

Factor's generic words are similar to Clojure's multimethods. However there
are times when protocols (in Clojure speak) are more appropriate
abstraction. Does Factor support a construct similar to protocols?

-- 
Cheers,
missingfaktor .

When you stand for what you believe in, you can change the world.
--
All the data continuously generated in your IT infrastructure 
contains a definitive record of customers, application performance, 
security threats, fraudulent activity, and more. Splunk takes this 
data and makes sense of it. IT sense. And common sense.
http://p.sf.net/sfu/splunk-novd2d___
Factor-talk mailing list
Factor-talk@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/factor-talk

Re: [Factor-talk] Shared libraries considered harmful

[L N]

http://www.forthos.org/
--
All the data continuously generated in your IT infrastructure 
contains a definitive record of customers, application performance, 
security threats, fraudulent activity, and more. Splunk takes this 
data and makes sense of it. IT sense. And common sense.
http://p.sf.net/sfu/splunk-novd2d___
Factor-talk mailing list
Factor-talk@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/factor-talk

Re: [Factor-talk] Shared libraries considered harmful

[Joe Groff]

On Thu, Nov 17, 2011 at 6:36 AM, Michael Clagett wrote:

> That brings a cost, however, as I've seen up to this point.  It
> distributes responsibility for documenting behavior across a community that
> like most communiteis is uneven in its fulfillment of responsibility.  The
> greater the dependency on "just read the code to figure out what's going
> on", the greater the barrier to entry to newcomers.  I have talked to more
> than one such person who has come to take a look and turned away for this
> very reason.  The power and flexibility is tremendous, no question.  But
> power and flexibility aren't always the only thing you need.
>

This is definitely a problem, and a similar problem afflicts most languages
that rely heavily on metaprogramming, such as Lisps and even C++. The
common substrate is too thin, and different users' code looks too different
to easily read or interoperate with. Factor at least benefits from having a
relatively small community and a well-centralized code base, so when we do
promote idioms and libraries to the core language it's easier for us to
propagate those benefits across all the bundled code. That said, being a
small community, we also lack the free hands to keep the documentation
polished and accessible to newcomers a lot of the time.


>  But wouldn't it be nice to have a platform that offers the power and
> flexibility of Factor that also plays nicely with OSes and other runtime
> environments?
>

I think Factor goes much further than most languages of its kind in
integrating with the host platform. Unlike Smalltalks, the canonical source
code isn't hidden in the image, and you can use your own text editor. More
OS features have bindings and bundled high-level APIs than your typical
Lisp or Scheme. The UI doesn't pretend to be a desktop in and of itself
(though granted, it still uses goofy non-native widgets). Deployed programs
aren't tens of megabytes in size. What do you feel is lacking in this
regard? Maybe I misunderstand your notion of integrating.

-Joe
--
All the data continuously generated in your IT infrastructure 
contains a definitive record of customers, application performance, 
security threats, fraudulent activity, and more. Splunk takes this 
data and makes sense of it. IT sense. And common sense.
http://p.sf.net/sfu/splunk-novd2d___
Factor-talk mailing list
Factor-talk@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/factor-talk

Re: [Factor-talk] Shared libraries considered harmful

[L N]

On Thu, Nov 17, 2011 at 9:36 AM, Michael Clagett wrote:

>  A couple thoughts --
>
> First of all, as far as I can tell Factor is one big shared library; so
> much of even the core language is implemented in library code that this is
> truer for Factor than it is for many languages one encounters.
>

I see.


>
> That brings a cost, however, as I've seen up to this point.  It
> distributes responsibility for documenting behavior across a community that
> like most communiteis is uneven in its fulfillment of responsibility.  The
> greater the dependency on "just read the code to figure out what's going
> on", the greater the barrier to entry to newcomers.  I have talked to more
> than one such person who has come to take a look and turned away for this
> very reason.  The power and flexibility is tremendous, no question.  But
> power and flexibility aren't always the only thing you need.
>
> As a development manager managing a group of about eight individuals, I
> have to say that one of the benefits of operating systems and runtimes like
> .NET and Java is the shared vocabulary they give a team to develop working
> practices and patterns.  The value of this obviously varies from situation
> to situation, but the greater difficulty that Factor and other flexible and
> extendible vocabulary-based development platforms can bring with them in
> terms of achieving standards and discipline is a concern that to me is very
> real and potentially limits the scope of their adoption.  Maybe it was
> never intended that Factor become a mainstream language suitable for use in
> primarily team-based environments and perhaps I am applying a standard to
> it that is not appropriate.  But wouldn't it be nice to have a platform
> that offers the power and flexibility of Factor that also plays nicely with
> OSes and other runtime environments?
>

The idea I had in mind is that there could be a system where the vocabulary
*is* the OS.

In this system, a program would just be a regular word in the vocabulary.

It is similar to the idea in TCL that "everything is a command".

Imagine in this system, there are two commands, "cat" and "more", like
their unix equivalents.

The "cat" command is composed of sub-routines, which are also regular words
in the vocabulary.

Assume that we don't want "more" to just be a wrapper of "cat".

We may want "more" to also be able to access the sub-routines of which
"cat" is composed.

This is easy, since the code for "more" can import the namespace of "cat",
and access its sub-routines.

I'm not sure whether "cat" and "more" actually share any code in normal
unix.

This is a naive example, of course.

I'm wondering, though, whether there is any reason why a complete OS could
not be written in Factor or Forth.

Also, I was thinking about how it is sort of fascinating how different
software realms are converging.

The similarity of Factor and Forth is interesting.

Is this a deliberate choice of design?  Is it also a sort of natural
inevitable convergence?

Is there a similarity between the Factor vocabulary and the apt-get package
system of Linux?

Both give you a giant list of stuff you can do with the computer.

It seems like a vocabulary is somehow more general than a package-list.

Then I wondered if a package-list could be replaced by a vocabulary.

Then the vocabulary would *be* the OS, and the listener would be the
command-line.

Someone pointed out to me before that this idea has long existed in Forth,
for special-purpose architectures.

I am just thinking out loud about doing the same thing on a modern PC.

  - Leonard
--
All the data continuously generated in your IT infrastructure 
contains a definitive record of customers, application performance, 
security threats, fraudulent activity, and more. Splunk takes this 
data and makes sense of it. IT sense. And common sense.
http://p.sf.net/sfu/splunk-novd2d___
Factor-talk mailing list
Factor-talk@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/factor-talk

Re: [Factor-talk] Shared libraries considered harmful

[Michael Clagett]

A couple thoughts --
 
First of all, as far as I can tell Factor is one big shared library; so much of 
even the core language is implemented in library code that this is truer for 
Factor than it is for many languages one encounters.
 
That brings a cost, however, as I've seen up to this point.  It distributes 
responsibility for documenting behavior across a community that like most 
communiteis is uneven in its fulfillment of responsibility.  The greater the 
dependency on "just read the code to figure out what's going on", the greater 
the barrier to entry to newcomers.  I have talked to more than one such person 
who has come to take a look and turned away for this very reason.  The power 
and flexibility is tremendous, no question.  But power and flexibility aren't 
always the only thing you need.
 
As a development manager managing a group of about eight individuals, I have to 
say that one of the benefits of operating systems and runtimes like .NET and 
Java is the shared vocabulary they give a team to develop working practices and 
patterns.  The value of this obviously varies from situation to situation, but 
the greater difficulty that Factor and other flexible and extendible 
vocabulary-based development platforms can bring with them in terms of 
achieving standards and discipline is a concern that to me is very real and 
potentially limits the scope of their adoption.  Maybe it was never intended 
that Factor become a mainstream language suitable for use in primarily 
team-based environments and perhaps I am applying a standard to it that is not 
appropriate.  But wouldn't it be nice to have a platform that offers the power 
and flexibility of Factor that also plays nicely with OSes and other runtime 
environments?
 
Curious what others think about this.
 



Date: Thu, 17 Nov 2011 06:45:00 -0500
From: leonardne...@gmail.com
To: factor-talk@lists.sourceforge.net
Subject: [Factor-talk] Shared libraries considered harmful

Does anybody ever dream about returning to the early days of Forth, where "the 
vocabulary" is "the system"?

http://www.colorforth.com/POL.htm

Does anybody ever dream about software that shares functionality with 
"procedure calls", rather than with OLE, SOAP, or .so files?

http://blogs.oracle.com/rvs/entry/what_does_dynamic_linking_and

Is it even necessary to package code into "programs"?

Perhaps "namespaces" provide more flexibility than "programs"?

Does anyone dream about the Factor vocabulary becoming so rich than a "modern" 
"OS" is unnecessary?

Cheers,
Leonard



-- 
All the data continuously generated in your IT infrastructure contains a 
definitive record of customers, application performance, security threats, 
fraudulent activity, and more. Splunk takes this data and makes sense of it. IT 
sense. And common sense. http://p.sf.net/sfu/splunk-novd2d
___ Factor-talk mailing list 
Factor-talk@lists.sourceforge.net 
https://lists.sourceforge.net/lists/listinfo/factor-talk
   --
All the data continuously generated in your IT infrastructure 
contains a definitive record of customers, application performance, 
security threats, fraudulent activity, and more. Splunk takes this 
data and makes sense of it. IT sense. And common sense.
http://p.sf.net/sfu/splunk-novd2d___
Factor-talk mailing list
Factor-talk@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/factor-talk

[Factor-talk] Shared libraries considered harmful

[L N]

Does anybody ever dream about returning to the early days of Forth, where
"the vocabulary" *is* "the system"?

http://www.colorforth.com/POL.htm

Does anybody ever dream about software that shares functionality with
"procedure calls", rather than with OLE, SOAP, or .so files?

http://blogs.oracle.com/rvs/entry/what_does_dynamic_linking_and

Is it even necessary to package code into "programs"?

Perhaps "namespaces" provide more flexibility than "programs"?

Does anyone dream about the Factor vocabulary becoming so rich than a
"modern" "OS" is *unnecessary*?

Cheers,
Leonard
--
All the data continuously generated in your IT infrastructure 
contains a definitive record of customers, application performance, 
security threats, fraudulent activity, and more. Splunk takes this 
data and makes sense of it. IT sense. And common sense.
http://p.sf.net/sfu/splunk-novd2d___
Factor-talk mailing list
Factor-talk@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/factor-talk