CVF 6.5A Debug Disimprovements

[Intel_C_Intel]

One month into using 6.5A, I had my first array overflow with the new compiler whilst running a Debug-compiled version of a program, but not running under Debug. If my memory serves me correctly (but I don?t take bets on it), in similar circumstances with 6.1A, one was offered the choice of Termination or Debug. Choosing the latter, one finished up at the offending line and was able to check the values of variables. 6.5A does not do this, but just notifies the offending line before expiring. In order to do what 6.1A did immediately, one needs to run the program again under debug until the error occurs. As far as 6.5A vs 6.1A is concerned, this seems to be a clear case of Hutber?s Law (Improvement = Deterioration).

A minor point is that, when stepping through a program under debug, the QuickWin Window flashes up briefly, although not always, but to a far greater extent than under 6.1A. Has this any connection with Alan Hathaway?s unanswered complaint (see ?Screen Flash? under 6.5A)? I too think that there is rather more flashing at the start of Quickwin programs than I remember, but I thought that mine always had a little flashing under 6.1A.

Another minor point with 6.5A is that when debugging, the focus sometimes no longer seems to be in the expected place, and I have found myself inserting new lines in the source code instead of stepping through the program, which rarely if at all happened on 6.1A. (Keep checking the focus, bear).

Yet another very minor point with 6.5A vs 6.1A is that although I am have upgraded my PC from 233MHz / Windows NT to 1.5GHz / Windows 2000, in the Development Environment Menu, the File drop-down box takes a very much longer time to display than before.

I am not going to reinstall 6.1A to check, but I?m not imagining all these points, am I?

Bear of little Brain

[sabalan]

I have just upgraded from 6.1A to 6.5A and it would be very frustrating to lose the first feature you write about. But I tested and it still works. Have you checked "Generate Traceback Information" (Project Settings | Fortran | Run Time)?

And your last point depends on nothing but the very W2K. You can see this phenomenon while running other programs, like Word and Excel, too.

Regards,
Sabalan.

[Steven_L_Intel1]

This is not a change in CVF, though something might have gotten reset. Go to Tools..Options..Debug and check the "Just in time debugging" box. I tend to leave this off.

Steve

[Intel_C_Intel]

Sabalan,

Many thanks for your reply - however, the "Generate Traceback Information" box was checked. On my other points, from your observation and from digging around, I think that Windows 2000 is probably the culprit in all of them, so here is yet another association of Hutber's law and Microsoft.

Steve,

Many thanks for your tip - yes, the "Just in time Debugging Box" was unchecked. As is usual in these matters, it is a mystery how it got unchecked. Other installations of CVF 6.5A here have all got the box checked, and I am fairly certain that is how it installed. A quick run shows that my installation is now offering me the option to debug when meeting an array overflow, so all is well .... until my next problem!

Bear of little brain