What you describe is what Lucid Chart used to do (still does, but is more complicated) and yEd (post 3) seems to do. I can take or leave the rubber banding -- IMHO it is less useful than the autorouter in Eagle 7.x.
What I reacted to and/or disagreed with was this:
Don't misunderstand me, if there were a program that could convert a flow chart like you describe into decent code, that would be great. But, I am not aware of one.
Can you give any examples of flow charting programs that meet that "should" criterion?
Hi,
No i was just throwing that out there for thought.
I can see how this can be very possible, but the reason i said "pseudo code" was simply because there are too many platforms to support, most likely.
Pseudo code could be very simple and allow you to easily convert it into your language choice. Some things would be the same or nearly the same.
For example, in the flow chart you type:
"A=B"
in a block and without the quotes, and the program spits out:
A=B;
It's then your job to define what types you want them to be or any other info like an initializer or something.
There would actually be two or more layers, and with the possible option to provide your own final layer.
For example, the first layer converts that block with A=B to A=B;, then your final layer converts that to:
float A,B;
A-B;
Of course there would be more to it than that.
I know this would work because i've done a similer IDE for Windows a long time ago that had to take in graphics objects and parse them, then convert them to Windows graphics objects according to the users specs. It would take in the graphics objects one by one, then prompt the user for what type it should be.
The flow chart program would know this already from the shape, or a shape and side note perhaps.
I've also managed to turn pseudo code into ASM for compilation, so i would not be too far from being able to code the whole program but it would take a lot of work. The ASM platform i used was for Microchip however, so there would be some changes for other output platforms.
The rubber banding idea was an idea for use with actually drawing the flow chart. When we add things, sometimes we have to move blocks around. If they held their connections, there would be no erasing and redrawing when a block was moved.
After reading my post over for typos, i realized that maybe we could use a circuit sim program for this if we can get an output node list. The output node list could tell us what *custom* objects are being used and how they are connected. The sim program would allow editing as usual so that would make this simpler. We'd have to be able to read the text too though.
I think most sim programs have a node list output option. I'll have to check LT Spice.
Unfortunately i dont think the text follows the objects like it should in a flow chart that is being edited.