Thank you for this information, Björn.

There are several steps needed and first of all break the task into
> smaller tasks.
>
> The first thing to note is that most operations can be done in several
> different ways and if you have a problem because one operation is
> taking long then you have to look at alternatives.
>

This is one thing I like about J: it seems to make it straightforward to
break a large task into smaller ones, and there are often many alternative
"idioms" which can be used to perform the same task.

>
> So everyone involved with the project need to adjust to the task at
> hand and learn it in steps.
>

I should clarify that I'm simply studying J by myself now - hoping
eventually to find a commercial application where I can use it.

>
> J can handle a huge number of data so that is not a problem and it can
> handle a lot of connections.
>

This is encouraging to hear. I'd be curious if anyone also has some specific
experience with using a J server handling a lot of connections. I don't
think I'd plan on using J for a public-facing web-server - but it would be
interesting to hear if people have been using any sort of J server in-house,
and how it performs when handling multiple connections.

I sometimes hear about thread-based versus process-based approaches for
handling multiple connections, and I'd be interested in hearing anyone's
opinions about how either of these approaches might work with a server
written in J.

I do not think J will be a problem but the learning curve of the
> people involved should be considered.
>

In this case, I only have my own learning curve to worry about :-)

I understand that J has not been very widely adopted, for whatever reasons -
but after looking at many languages, I keep coming back to languages like J
because they seem to make the process of programming much more enjoyable. I
think my life might be rather dreary if I had to work in C++ or Java or even
C# or Scala all day long.

So I'm considering devoting a good-sized chunk of time to really learning J.
I studied K for a few months a while back, so I'm already starting to get a
good grasp of concepts like loopless programming and adverbs. And while J
might not be as strong as K in terms of raw number-crunching ability, it
seems to come close, and J602's ability to include OCX controls gives me
hope that J could be used to create decent (if not spectacular) GUIs.

I'm hearing from other people on this thread (for example, Alex Rufon) that
J might not be very good at transaction-oriented or record-based processing,
and I'm still trying to find out whether that's an inherent, insurmountable
lack in J, or whether capabilities for concurrent multi-user
transaction-oriented / records-based processing and editing could be added
to J simply by implementing some of the existing concurrency solutions out
there (such as Firebird's multi-version concurrency control, or the more
advance and generalized Paxos algorithms).

>
> Once everyone is up to speed I do not see any problem you should not
> be able to do what you want.
>

Thanks for your helpful encouragement!

- Scott
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to