Richard Harter's World
Site map
April 2011
email

Letters to the editor, April 2011

This a traditional letter column. You are encouraged to write a letter of comment on anything that you find worthy of comment. It will (may) be published in this column along with my reply. As editor I reserve the right to delete material; however I will not alter the undeleted material. E-mail to me that solely references the contents of this site will be assumed to be publishable mail. All other e-mail is assumed to be private. And, of course, anything marked not for publication is not for publication. Oh yes, letters of appreciation for the scholarly resources provided by this site will be handled very discreetly. This page contains the correspondence for April 2011.

Index of contributors

Other Correspondence Pages


From: Tim DeLaney
Date: 12 March 2011
Subj: Shameless self-promotion

I've been published by the Secular Web: Christianity: An Irreverent Overview http://www.infidels.org/kiosk/article830.html I'm sending this to all my friends and relatives who might be interested. Tim

My apologies for not replying sooner. Your article is well written albeit a bit long. The problem is that it is obvious that the Christian narratives are grossly improbable and shot through with contradictions. Clearly this doesn't bother believers. The secret is that religion is about Truth and that Truth doesn't have to be factual or make sense.
Return to index of contributors

From: Mark Taylor
Date: 7 March 2011
Subj: SAN programming language

Many thanks for your reply & your offer to look at my ideas & comment on them - this is very welcome to me, as you are the first person to offer who has any real practical experience.

You are welcome.
http://lambda-the-ultimate.org/node/4075

which is a brief explanation of what I am thinking, inviting comments and/or collaboration.

Some of the comments were very interesting.
I had written
As it happens, I didn't build a working implementation - at least not yet. From your perspective, that's the good news.
This is actually not good news to me! I had been hoping to learn from other people's work!
I didn't build a working implementation of the SAN language; however I did build a working implementation of the data flow engine. In the course of writing it and rewriting it I created a series of test programs. These illuminated my mind. [snip]
I would be more than happy to look at your ideas and offer some commentary. However, if I make a small suggestion, it would be good to sign on to the google group, flow based programming, and post about your ideas there.
I am already signed up, and the post I referred to above on LtU, is also there. They are an enthusiastic and relatively active crowd, but I have not been able to interest them in the idea that a new language is needed - maybe that was unrealistic of me.
It's probably a matter of paradigm burnout.
I had written
I have become persuaded that flow based programming requires a "two language" approach. That is, a program or application consists of a description language that describes the connectivity of the flow based network, and the language(s) in which the components are written. In fact, one may really want a third language, a data description language.
I have come to the almost the same conclusions.

In my scheme, what will be flowing along the pipes (connections) will be data objects, and to achieve this, a data definition language will be essential. I've yet to start thinking about this though.

However, I have devised an outline scheme for the first two languages. And I was keen to make the first (the flow or connection language) as transparent as possible. All the implementations I have seen so far (including Paul Morrisons) have arcane & complicated features like ports, locks, synchronisation issues, deadlock issues and other complications. I thought that these really should not have to be dealt with by the application programmer, otherwise all that has been done is to exchange one set of complications for a different set. I was therefore desperate to make my "connection" language as simple as possible, or even better to make it 100% transparent - i.e. no connection coding has to be done at all, so that the requirement for a "connection language" *disappears completely*. And recently, I think I might have managed to do this.

I'm not 100% certain because the limitations imposed by my scheme, might make it impossible or very difficult to code some algorithms or solve some of programming problems. I'm still thinking this through & hence the jury is still out.

I would be interested in hearing what you came up with. I went through several iterations in my thinking on this issue. I started with port numbers. I switched to port identifiers. Eventually I moved to a type system which may seem elaborate at first but actually works quite smoothly.

The key idea is that messages (objects if you like) have a type label. Inports and outports have a two part type label. Part 1 is a message type; part 2 is a port number. The port number is only used if there is a need to distinguish between auton (component) ports with the same message type.

Connection types are defined as an (outport type, inport type) pair where message types must match. Actual connections are specified as an (auton, connection type, auton) triplet. Thus:

	Alpha -> (integer) -> Beta
A key point of this scheme is that one can build up a library of connection types. Furthermore it ensures type consistency in connections.
I am also aiming to make my text-based language correspond directly with a graphical (or diagrammatic form), so that programming can be done either textually or graphically, and the source code can be easily & unambiguously translated between the two forms.

For my component language, I have pencilled in a small "standard" control-flow language that would probably look like Pascal or Modula-2. But I am not ready to this decision yet, because one of my aims is to make my system general purpose and low-level, such it can be self-compiling. And to do this I will have to design the component language carefully.

You might want to think about that a bit. Components could be almost any language provided it supports the interface conventions. The question of interest is whether the system conventions can be enforced.
I had written
Be that as it may, the dataflow engine code is almost ready for prime time. I expect that I will go back and revise the San specification yet again, and then resume implementation.
I will be very interested to see your work when its ready.

Can I ask... do you do this in your spare time as a hobby? And do you have any experience designing programming languages? Although I work at a University, I am not on the academic side, and my interest is not directly related to my profressional IT work.

This is definitely a spare time proposition. I have some experience at designing and implementing languages, albeit scripting languages. This was subordinate to main occupation for many years, which was software configuration management systems.

The big thing that I need to do in passing the engine off is to document the interface and the theory of usage.

Return to index of contributors

From: allen percy
Date: 28 March 2011
Subj: Richard Harter

Greetings, Richard Harter

we have completely new and interesting position for you.

Top-rated European firm searches for United States partners to help in United States financial deals and execute local representative responsibilities. This position takes zero start-up Investment, or past work experience, we provide you with full training and personal assistance.

Our company gives ability to get up to $15K a month with 3 hours spent during the week. That work can become another source of inflow or basic job as it is {hb} job. Please, send REPLY and our staff will email you next portion of info kit very soon.

Wishing you the best.

Gentlemen,

I do wish to thank you for thinking of me. I certainly could use a new source of income, particularly since my demotion from Chief Bottle Washer to Assistant Bottle Washer was accompanied by a sharp salary cut. Unfortunately I must decline your offer. My experience with {hb} jobs has been quite unfortunate - my physician keeps muttering about there aren't any more new antibiotics to try.

Please do keep me in mind when next you have an immense lottery prize to give away.

Return to index of contributors

From: Nathan Childers
Date: 11 April 2011
Subj: Death in the Garden of Reason

My dear boy, I must say it is a pleasure to read your review. It was all that I have come to expect from you. Given that you saw fit to provide a plot summary (I thought we had broken you of that habit) I will say that your summary was ingenious - although technically correct it to omits all of the important elements in the plot. I dare say anyone who reads my book on the strength of your review will be quite surprised. By the bye, I thought your little confession - "To be honest, it is hard to know what to make of this book" - was quite touching. Would that more reviewers were as honest.

Thank you. So many authors curse their reviews. I look forward to reading whatever you might write next.
Return to index of contributors


This page was last updated April 11, 2011.

Richard Harter's World
Site map
April 2011
email