Assignment to unallocated derived type / array

[lklawrie1]

This one surprised me. (From another developer using a different compiler). I verified it using CVF -- have not tried IVF yet.

Derived type is not allocated due to various reasons but the assignment:

x%y=0

flies by okay in CVF. No error.

Any idea what compiler flags I might use to make sure this trips up?

Linda

[Steven_L_Intel1]

There isn't anything in CVF, or really IVF, to catch this at this time. It's on our list. You might get an array bounds error in IVF if you had that enabled and tried it, or maybe an access violation.

This is a general class of pointer and shape checking that is missing from the product.

[lklawrie1]

FYI -- I guess g95 has it as this is where it failed. OTOH, they don't have debugging or apparently very good traceback so he (the person using the source code) had to resort to putting in print statements for tracing the problem. :-)

Thanks. I just tried IVF -- inmy release version (which doesn't have extensive array bounds tracking turned on) it doesn't trip.

Linda

[Steven_L_Intel1]

Yeah, this is a sore point with me - we really need to do a better job with run-time diagnostics. The more people who file feature requests for such things, the more likely they'll get done.

[lklawrie1]

Do I need to do something more (to file a feature request)? If so, let me know.

Linda

[Steven_L_Intel1]

Yes - file an issue with Intel Premier Support with the report type of Feature Request.

[lklawrie1]

Will they want an example file for this one?

Linda

[Steven_L_Intel1]

It would certainly help.