Thanks for the response.

The article you pointed is a bit complex for me as I lack knowledge of ‘formal logic’ and I’ll have to gain significant knowledge in ‘type theory’ and other things the article is based on or referenes, before I’ll be able to understand and appreciate the concept fully and be able to actually put it to use. My background is IT and not Computer Science.

I got exposed to functional programming through one of your java’s ‘throws’ Monad article.
It got me interested enough to research about Monad.
And now I want more

Since you got me in this confusion in the first place, I would like you to to help me out a little

IT would be very helpful for me and newbies like me if you could write up an article on how to use FJ to create an actor-based server application with some example code.

Armed with an example I understand, it would help me explore more into functional programming and use it to solve problems better if it permits.

Thanks for the response.

I had great pleasure reading your comments and was a bit disappointed when it came to an end though

And also I was a bit disappointed to realise that out of the two programmers camps you mentioned, I unfortunately fall into the second one.
Although you mentioned that its an objective refutable fact - and I take your words for it,
I’m keeping my chins up with the help of my assumption that no empirical evidence has yet been found that says the ones from the second camp cannot cross-over to first one over time.

I’ll keep my eyes peeled on your blog due to learning opportunity available here ;).

I’ll see what I can do regarding echo server. Keep an eye out

I followed a few FJ tutorials and found them very impressive.
But being a mere application developer with just enough understanding of Java to make it work mostly, I found the examples a little hard to grasp and replicate them in a meaningful way for me.
I build a Java TCP server for a vehicle tracking system and I was pleased with my work as it was my first attempt at socket programming as well as multi-threading and Its been working quite well apart from GC problems. But due to a new business prospect at our company I need to quickly come up with a better solution than the one I have currently, as the number of connections to the server is likely to increase by anything between a few hundred to 1-2 thousands per month.
So my current challange is to find a good vertical scaling solution before horizontal scaling so the hardware cost and maintenance/operation complexity is kept at mimimum.
I have looked at few alternatives, learn Erlang, learn Scala or other dynamic languages that run on JVM that allows me to write Actor style application, or find a solution within Java.
Since I like the third option for obvious reasons, I have looked at a couple of options.
Kilim was the first one I looked at besides others.

I like the idea behind Kilim but I’m not very comfortable with the idea of byte-code manipulation.
I’m a big fan of unobtrusive drop-in-libraries and FunctionalJava is very appealing to me in that matter.
But I find the examples a bit hard to grasp.

So, is there any chance that you could write up a quick and dirty Echo Server like example that utilises Actors to do some trivial tasks?

It would me immensely appreciated and I’m sure it will be beneficial to other people looking for alternatives to the current shared-memory thread programming.

What I can say is that I have much less angst over reasoning about dynamically typed languages than you appear to have. I respect your right to your preferences but don’t really see the big deal. All the same types are still there, you may just have to infer them or apply higher order reasoning. I really don’t see what the problem is.

For me the really hard systems to reason about are the large complex ones. Functional flavoured langages and techniques give me one set of tools for finding ways to reason about smaller parts of larger systems. Imperative and constraint-based languages give me other sets of useful tools and I don’t claim that the best tools for reasoning about such languages are lambda calculus based - indeed if I tried to do that I would probably be dismayed at such languages - so I don’t.

I do agree with your comments about unit tests being theoretically undesirable in some scenarios but exemplar driven steps within your development approach are sometimes the most crucial ones as they allow you to validate your theorems and code with your customers conceptual models of your theorems and code.

Let me make something very clear; all programmers fall into one of two distinct camps:
1) Those who write logical theorems and know it
2) Those who write logical theorems and do not know it (and suffer significant adverse consequences).
This is an objective irrefutable fact that cannot be false no matter how hard you and I wish it to be so.

I’m well aware of many other logically consistent ways of observing programming and I use them all very effectively. I am also aware of the many pseudo-scientific memes that perpetuate in our industry, especially in dynamic languages (yes, even one as desperately lacking as Groovy). Indeed, I am more than capable of degenerating into a mindset that is perpetuated by these non-ideas, however, I elect not to, not out of fundamentalism, but simply because it would not be useful.

It is remarkable that my earlier analogy to Creationism (out of the many possible) was followed by an argument that is often used by Creationists! Your comment boils down to “why limit oneself to an understanding of reality, when there are so many delusions of grandeur available to indulge?”. It is as if you are advocating that I accept some alternative to reality, on the assumption that I do not understand that alternative. Let me also make this clear, I can use your dynamic language as effectively as you can, which is not very effective at all. I simply elect not to, because it offers nothing in my interest (or yours for that matter, but you don’t know that) at a significant disadvantage.

I don’t wish to sound mean and I hesitated before deciding to say what clearly needs to be said, but it would really pay for you to know what you are talking about. I say this simply because there is quite possibly a learning opportunity available. Also, it would allow this discussion to continue without silly remarks being made.

I am not bemused by your desire to limit yourself to wilful ignorance, simply because my armchair cognitive science knowledge explains why it is so, but this is another matter (George Orwell calls it double-think; I call it Cognitive Dissonance). I think discussion is going nowhere due to silly remarks (Creationists, take note), but to turn it into something meaningful, it might pay to start down a path of education. This would require an admission by yourself, even if only to yourself.

Are you aware that the untyped lambda calculus is an instance of the typed lambda calculi? Are you aware that with sufficient utilisation of a type system and for some programs, the need for unit tests disappears completely (in that, it is not only not desirable to write a unit test, but impossible because there is nothing more to say)?

Alternatively you might be saying that writing code is not proving a theorem, in which case, I advise you to prepare yourself for your Nobel Prize I suspect this is not happening mostly due to the audaciousness of the claim.

Please clarify the misunderstanding. Like I said, I am not going to debate because there is no debate to be had (admittedly, this is slightly a lie, but I’m taking a Socratic Midwife approach here).