Debugging support includes decent error messages, and being able to
use the command line interactively, but yes being able to stop and
inspect values and stack traces can help quite a lot also.

The last time I used J.dll in a windows program that I was compiling,
most of my time got burned on trying to figure out what was triggering
errors (though, admittedly, a lot of that was not on my machine but on
the machine of the guy I gave the program to - and a lot of those
errors had to do with file permissions on networked files - these were
problems that only showed up for him).

This time, though, the issue is that I'm wanting to do some
exploratory programming - delving into algorithms where I have only a
foggy idea of how I'm going to get it done.

The other issue, though, is that testing the algorithm for accuracy
mostly means inspecting visual results. That's why I'm a bit foggy
right now - I've gotten as far as I can go dealing with simple cases
an I need to see how things work against "real data".

Like, for instance, one aspect of the problem has to do with efficient
partitioning of a 3d space (into however many thousands of
polyhedrons) but in a J friendly manner (kd-trees aren't something I'd
want to use within J, and while calling out to a foreign
implementation might work I also want to play with some ideas that I
have for alternative algorithms) but another aspect has to do with
picking visually pleasing heuristics.

Anyways, this is ill-defined-enough that I can't just sit down and
write the code without having some kind of test rig that helps me see
what I'm doing - several senses of that sort of concept.

But I guess perhaps what I should do is incorporate the J engine as a
project and use visual studio as my J debugging environment as well.
That might work...

Thanks,

-- 
Raul

On Sun, Feb 15, 2015 at 9:52 AM, Joe Bogner <[email protected]> wrote:
> Ok, that makes sense. I don't have any experience with the J debugger so I
> can't comment on it. My gut is to replicate the the functionality in your
> program instead of calling the IDE. It seems the debugging core is
> implemented in the engine and the IDE controls it. On the surface, it looks
> like it would be less work to just load debug.ijs and wire up your own UI
> to it. I am only speculating though
>
> http://www.jsoftware.com/help/user/lib_debug.htm
>
>  On Feb 15, 2015 9:09 AM, "Raul Miller" <[email protected]> wrote:
>
>> I very much want the J IDE for the J side of things, for my initial
>> explorations. The reason is J debugging support.
>>
>> Meanwhile, the other program uses wxWidgets (and needs elevated
>> privileges for some reason - I think it's because of rendering, but
>> it's just using opengl so I'm actually not completely sure why... I
>> should look into that).
>>
>> I don't think I can integrate Qt and wxWidgets - both want to own the
>> top level of control for the program. So I just want some way for the
>> two to communicate with each other. So perhaps I am stuck with using
>> files...
>>
>> That said, if there's a way of using JHS with j.dll wired into another
>> program, that would work for me.
>>
>> Thanks,
>>
>> --
>> Raul
>>
>>
>> On Sun, Feb 15, 2015 at 8:46 AM, Joe Bogner <[email protected]> wrote:
>> > Do you need the IDE or the ability to call J primitives and scripts on
>> the
>> > data? The IDE would be the REPL, the project manager and wd functions to
>> > build forms in QT. The latter would be the J engine
>> >
>> > I think it would be unusual to integrate the full IDE into another
>> program.
>> > You would probably need to start with the QT IDE source and make sure you
>> > are compliant with licensing
>> >
>> > To follow the more common case of integrating with the J engine, that
>> link
>> > is a good guide or there is an example at
>> > http://jsoftware.com/wsvn/addons/trunk/general/misc/jdll.ijs
>> >
>> > On Feb 15, 2015 8:20 AM, "Raul Miller" <[email protected]> wrote:
>> >
>> >> What's the best way to wire the J IDE into another program?
>> >>
>> >> In this particular case, I've got the source to the other program
>> >> (it's C++). I'm running on windows (windows 7), and I can run the
>> >> other program in visual studio express debug mode. It's got some
>> >> decent sized matrices (thousands of rows of geometrical data) which
>> >> I'd like to bring into J to manipulate an then toss back to the other
>> >> program (for it's user interface).
>> >>
>> >> Is there a way of using the mechanisms described at
>> >> http://www.jsoftware.com/help/user/calling_jdll.htm for this kind of
>> >> thing? Or do I need to resort to writing and reading files?
>> >>
>> >> Thanks,
>> >>
>> >> --
>> >> Raul
>> >> ----------------------------------------------------------------------
>> >> 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
>>
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to