[Date Index] [Thread Index] [Author Index]

Re: Using Workbench to debug Compile[] ??

• To: mathgroup at smc.vnet.net
• Subject: [mg127567] Re: Using Workbench to debug Compile[] ??
• From: David Bailey <dave at removedbailey.co.uk>
• Date: Sat, 4 Aug 2012 21:45:34 -0400 (EDT)
• Delivered-to: l-mathgroup@mail-archive0.wolfram.com
• Delivered-to: l-mathgroup@wolfram.com
• Delivered-to: mathgroup-newout@smc.vnet.net
• Delivered-to: mathgroup-newsend@smc.vnet.net
• References: <jvdffu$c5t$1@smc.vnet.net> <jvg1k4$rkl$1@smc.vnet.net> <jvis3r$2uf$1@smc.vnet.net>

On 04/08/2012 11:04, Axel wrote:
> No, this is just a general question.
> I want to know if it is at all possible to debug a function that is returned by the Compile command with the Workbench. As I see it, this is NOT possible because I now have bytecode and not any more Mathematica source code.
> I just would like to confirm this.
>
> If you know any other way to debug compiled functions please let me know.
>
> Thanx
>    Axel
>

Well I haven't used compiled code explicitly much, but as I understand 
it, the results should be the same as non-compiled code, except perhaps 
for rounding errors.

Therefore, if I had code that worked in interpreted mode, but not in 
compiled mode, I would try to thin the code down to the smallest 
fragment that showed a problem, and then perhaps report it as a bug to 
Wolfram.

I would also test code before trying to compile it - so if you are 
talking about bugs that are there anyway, get them out before you use 
the compiler! To do this, you may need to scale the problem down a bit 
so that it can execute without the compiler in a reasonable amount of 
time. Usually this is fairly easy.

David Bailey
http://www.dbaileyconsultancy.co.uk