There's nothing specific to databases about the value of J. I've never
done anything with databases before my current project. I've used J to
model the transformation pipeline of a graphics processor, the
texture-mapping system of a flight simulator, a stock-trading program, a
program to play Connect Four, a lot of code for recording and
manipulating MP3 files, a fair amount of image processing... The only
project I ever had that I considered J ill-suited for was parsing MIDI
files, and that was pretty early in my J life & I could do it better now.
Henry RIch
On 11/28/2017 4:38 PM, Andrew Dabrowski wrote:
Thanks. I keep forgetting that J and its emulators are really popular
for databases. That's not something I deal with, so perhaps that's
why I'm blind to some of its virtues.
On 11/28/2017 04:21 PM, Henry Rich wrote:
I use J for all my own work because I get results fast. Example: I am
currently working on a database project, a big key/value store
written as about 200KLOC of C++, and I proposed a major design
change. Management was supportive but concerned, so I offered to
write a simulation of the system including the changes. I said I
could do it in 8 days, and using J, I did. Well worth doing, too, as
it led to refinements in the design. I don't know how long it would
have taken in C/C++, but I would be thinking months rather than days.
The combination of interpretive execution, terseness of expression,
and array-level thinking makes me more productive using J than I've
ever been in a scalar language. I completely disagree with you about
languages being equally good. When it comes to getting a program up
and running quickly, J has an edge in most of the places I've used it.
Yes, but for rapid development isn't Python (or Mathematica) just as
good if not better?
J is a language for describing a computation. C/C++ is a language
for telling a computer how to execute a computation.
I like that distinction. But J seems to get bogged down in syntactic
issues. As a beginner I find it impossible to parse a moderately
sized tacit expression. No doubt one gets better at this, but like
all computer languages, the one dimensional space it lives in seems to
confound any attempts to represent mathematical ideas directly.
A computer language based on mathematical notation sounds like a cool
but impractical idea. It would to have to be 2 dimensional, as in
fact math notation is.
If you don't need to focus on the execution details - that is, if you
can take your head out from under the hood and think only about what
needs to be done - you can save a lot of time and effort by staying
at the higher level. You have to train yourself to do that, though,
and doing so is harder than you would expect.
Henry Rich
On 11/28/2017 3:59 PM, Andrew Dabrowski wrote:
As much as I've complained about J in these forums I've been having
a good time translating some simple code into J. Someone gave me
wise advice, to stick with explicit definitions until I know the
language well, which advice I have cordially ignored because I'm
having too much fun playing code golf with tacit tangles.
I was fascinated by J because it seemed to try to build on aspects
of the human linguistic system. Natural language unfolds in one
dimension, time, so everything relevant to understanding a
particular word in a sentence either came before it or is yet to
come. J seemed to emulate this by having verbs which relate
directly only to objects on the immediate left and immediate right.
Moreover J seemed to be following a linguistic paradigm in have
nouns which are inert, verbs that act on nouns, and adverbs which
modify objects. This seemed like a promising way to exploit humans'
natural linguistic capabilities.
But maybe that's not way the J community currently sees J. Do you
love J most because of (pick only one)
1. the NL inspired syntax;
2. the suite of array utilities;
3. the concision of J code;
4. its being open-source; or
5. _____________________?
I've come to feel that all programming languages are ugly
compromises that are about equally good/bad at solving practical
problems, and the "best" language is just the one you know the
best. I used to be contemptuous of Perl, but after having learned
it well enough for my purposes I now kind of enjoy the brain teaser
quality of trying to fit problems into its procrustean bed (although
I still think it's a silly language). I have no doubt that I could
live happily with J as my primary language, but only after an
extended period of being handcuffed to it and forced to assimilate
its quirks. I don't know that I'll have the patience for that.
Is there any project in the J repos that demonstrates the strength
of J, as opposed to just showing that it's at least as good as other
languages? Any project that would have been significantly harder to
complete with say Python? Does J have any killer advantage, even in
just one aspect of programming? Or does J just appeal to you the way
pistachio ice-cream might, it just tickles your palate in a
no-accounting-for-taste way? That's how it appeals to me.
I was hoping someone could talk me into studying J seriously, but
now it looks to me like a language which, with APL, has had enormous
beneficial influence on many other languages, but which has failed
to learn in its turn from them. J seems a tad solipsistic.
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
---
This email has been checked for viruses by AVG.
http://www.avg.com
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
