IVF 8.1.021 has problem?

[lklawrie1]

Is there a link for release notes for the new IVF 8.1.021 compiler? The release notes on disk appear to point to July 2004.

My program no longer builds successfully. Gets through the compile (though something that was previously a warning is now an error) and then says

"Cannot open C:My Documents" argument list (or something like that).

Can you tell me what's wrong?

Linda

[Steven_L_Intel1]

The release notes don't have a date as far as I know. But there haven't been any changes to the release notes since the 8.1 release.

I have seen one other user report the "cannot open" problem. Please report it to Premier Support, but in the meantime, try changing your TEMP and TMP environment variables (Right click on My Computer, select properties, advanced, Environment Variables) to a valid path that does not contain a space.

I don't see this problem and my temp path includes a space, so I don't know why you are seeing it. If you'll upload the buildlog.txt from your build when you report the problem, that will help.

[Steven_L_Intel1]

Ok, I now see what happened. There was a fix made to prevent the problems people had when building projects with so many objects that it overflowed the link line. Unfortunately, the LINK command generated fails to quote the path to the "response file", so if that file, which is in your project, has a path with a blank in its name, you get this error.

The workaround for now is to move the project to a folder whose name does not contain blanks. We'll fix this. Sorry for the inconvenience.

[lklawrie1]

Thanks, Steve. I probably never would have seen the problem if I hadn't been lazy and accepted the default place where the projects get put. It's a bit silly to put them under "My Documents" in my mind but I understand the reasoning.

I will try that a bit later (went back to prior version) and I uploaded a couple more files that will let you see the other fortran problem. (to explain a bit more there -- it is intended to work if you compile single precision as put in the source code, but when you compile double precision, I think the compiler gets a bit confused. Maybe I should just have a conditional compile line or something as some other compiler might complain as 8.1.021 now is.

Linda

[Steven_L_Intel1]

Ok, I'll look at it when I get a chance. I'm travelling next week but will connect when I can. I did compile your code with /real_size:64 and it complained as I expected it would, since you had two generic routines that were then ambiguous. Maybe you're referring to something else....

[lklawrie1]

The ambiguity is what I was referring to. In 8.1.019 there is no warning or error (maybe I have warnings turned off). In 8.1.021 it's an error that can't be gotten around.

Linda

[Steven_L_Intel1]

Ah - well in that case 021 is finding a bug in your code that 019 didn't. If you tell it that REAL is double and don't then say that DOUBLE PRECISION is quad, then you have two specific procedures with the same actual signature, even though the datatypes are spelled differently.

Perhaps specifying /double_size:128 will do the trick for you? It allowed it to compile for me. If you don't use the DOUBLE PRECISION type then it shouldn't matter.

[lklawrie1]

Have verified that my project will build properly in 8.1.021 when there is no space in the path name.

Linda

[Steven_L_Intel1]

The link bug has been fixed, and I think the fix will appear in the next update.