Yonatan Zunger (zunger) wrote,
Yonatan Zunger
zunger

rgh...

Code runs. Code reads in data. Code runs elaborate curve fits, parametric models, calculates useful parameters. Code then writes the input parameters to the output file, instead of the output data.

Of course, I only notice this after the code runs.

Sigh.
  • Post a new comment

    Error

    switch
    default userpic

    Your reply will be screened

    Your IP address will be recorded 

    Help
  • 3 comments

woody77

August 7 2003, 17:38:11 UTC 12 years ago

That's ok, I've discovered that my compiler thinks that I have two variables with the same name. It's quite fun, too.

bool made_repairs = false;

{
//nested 3 ifs deep:
made_repairs = true;
}

return made_repairs;

The address of the variable changes between setting it true, and the return. I don't get it.

Aren't computers fun!?

zunger

August 7 2003, 17:56:25 UTC 12 years ago

Could it be a scoping issue? You might have accidentally stuck a "bool made_repairs" somewhere inside one of the if's, so the compiler defined a new symbol with the same name in that limited scope, and set it to true instead of the original.

(Yes, I've been doing a lot of C++ lately. Does it show? :)

woody77

August 8 2003, 14:26:57 UTC 12 years ago

No, it's not that simple (I've already checked that). But it's a template function, instantiated multiple times within the same calling function, and I really don't trust the MS DevStudio C++ compiler any farther than I can throw it.

And I don't know intel assembly well enough to go into the assembly, and see what the hell the compiler actually did.