From: Joseph M. Newcomer on
On Wed, 23 Jun 2010 15:22:51 -0700, JRGlide <JRGlide(a)discussions.microsoft.com> wrote:

>Thank you everyone for your answers. I have lots to think about now, but at
>least I know it can be done, which is the first hurdle.
>
>As for Joe�s question about unloading the DLL, it is partly a MATLAB issue
>and partly my design flaw. Here is the data flow:
>
>
>*************** ************** *************
>* * * * *
> *
>* MATLAB * --> * mex function * --> * 3D Viewer *
>* * * * *
> *
>*************** ************** *************
>
>Since the mex function is a DLL, MATLAB will not return until the mex
>function returns. But I also implemented the viewer as another DLL, so the
>mex function does not return until the viewer also returns. But if I convert
>the viewer to an exe and implement the handshake people have suggested, then
>I hope that it will stay resident.
>
>As far as ActiveX as Joe suggested, that was going to be a future question!
>I read about it once and it didn't sound easy. MATLAB supports ActiveX
>controls, which might be nice within a MATLAB GUI.
****
ActiveX is not trivial.
****
>
>As for the data format, the data is simple (x,y,z) floating point data so
>the data passing is simple. However, I missed my memory estimates by about a
>factor of 10! I can't see anyone wanting to pass in more than a million
>points, which is 36 MB of data, so it shouldn't be in the 100's like I said.
>Then again, no matter what capabilities this has, the engineers always want
>more!
****
Looking at the other considerations, the notion of a memory-mapped file looks like the
best option.
joe
****
Joseph M. Newcomer [MVP]
email: newcomer(a)flounder.com
Web: http://www.flounder.com
MVP Tips: http://www.flounder.com/mvp_tips.htm