ignoring unknown option "-Qtrapuv"

[lklawrie1]

I just tried a couple of different switches in my compile options.

including "set local variables to NaN" and am now getting the subject warning in my compile.

What gives?

Linda

[Steven_L_Intel1]

You've got the 9.0 IDE integration installed, but are using the 8.1 compiler. That option is new in 9.0. Also, the option doesn't do what that description says, and the description will change for the final 9.0 release.

Message Edited by sblionel on 05-12-2005 11:50 AM

[lklawrie1]

This apparently goes back to that support issue about integration problems.

So, "set stuff to NaN" is not viable?

Sorry for any confusion. (I should be the only one confused)

Linda

[Steven_L_Intel1]

If you recall, you were somehow stuck with the 9.0 integration even though you thought you uninstalled it. I had responded about a month ago with a suggestion for how to resolve it, but did not hear back from you.

Even in 9.0, this option simply sets stack-local values to an unusual value, but not a NaN.

[lklawrie1]

I haven't had a chance to try what you suggested. (Was in the middle of a release -- where I had to get things compiled and out). I will try today or tomorrow and see if it does resolve that issue.

"Just repeating what the option said -- that it would set local variables to NaN".

Linda

[Steven_L_Intel1]

I know. The description in the IDE is wrong (and it looks as if it's still going to be wrong in 9.0 - I'll look into that.) What it does is set values to hex CCCCCCCC, which is not a NaN. The manual now says "Initializes stack local variables to an unusual value that may help detect uninitialized variables."

[lklawrie1]

Hmm. I wonder if it actually will. (help find uninitialized variables)

I think it was the CDC compiler, many, many years ago that could actually set "core" to "indef" -- or maybe that was their linker/loader. But then, if I remember correctly, they had a specific bit set that was "indefinite".

two steps forward, 1 step back.

Linda