Converge VM

There is news at the Converge PL site to state that

A new RPython-based VM is now a drop-in replacement for most tasks, and is 3-10x faster than the old VM.

which can be seen over at the Converge GitHub site.

Builtin Functions

noted: Object icon's 76 builtin functions updated documentation at code.google

Object Icon updates

There are  Version 2.5 updates to three topics at the code.google Object Icon wiki.
These are

1. co-expressions
2. built-in functions
3. implementation notes

Btw, for my Cygwin build I needed to alias gawk as awk with an s-link and it happened that I had to start xwin with

/usr/bin/startxwin -- -nolock

due to some update glitch with Cygwin.
A run of

make test

gave one error which is likely a Cygwin issue on file permissions.

The Icon Visual Interface Builder, ivib, ran fine and my own test app tested OK. Remember to have at least one button with an event that you can code as exit().

Unicon article

Here is a rare article on Unicon.

Object Icon

Object Icon 2.5 is now available at code.google.com

Object Icon co-expressions: name, succeed, fail and fate

Robert Parlett has released Object Icon 2.4 as an O-O UNICODE implementation of the Icon programming language.

But another Objecticon is also appearing as the trademark of an unrelated dot com.

This is sad, given the much earlier choice of Unicon for extended-Icon.

Was ICON a sorrier name choice than SNOBOL ?

IconOO appears to be available: "Icon-Oh-Oh" ?

IconCR (as in Icon co-routines) also appears to be a trademark (Eye-Conquer.)

What remains is the sobering reality that UNICODE came to so many advanced languages so late.  How and why and where did "innovation" fail to occur in this area?

Then there is Curl from www.curl.com which was UNICODE from the start and never got a foothold.  And failed to protect its MIT trademark ...

Is it possible for "Object Icon" to be protected?

ObjectIcon HttpServer without minify-me

Here are a few notes on code in ObjectIcon 2.4 httpclient.icn and related files.

class HttpClient(NetClient)

# and in netclient.icn

class NetClient(Connectable, SetFields)

# in connectable.icn
package util

class Connectable()

My first thought is that Connectable - with no parents, no imports, would have served so much better as a Trait in a traits package. Here is the top of setfields.icn

package util

import lang

With that import, would SettableFields be a Trait?  Here is the head of that class

class SetFields()
   #
   # Generate the {Field}s for the given list.  Note that the list is
   # emptied by this function.
   #
   public static generate_fields(l)

Field is a class found in the same file.  But the problem lies in the last method of generate_fields with

suspend Field(attr, val)

Now we do not have a trait.  And then there is non-documenting code  (really not Icon-esque in 2010 ) which should at least have been

public static generate_fields( lst )

because this is not minify-me JavaScript at Google or Yahoo or qtask.com or...

This is not some stylistic quibble: in open-source code this can be decisive.  Style guidelines count as a matter of quality code.  Quality does matter.