This week I've finally managed to find time to test the new compiler. I know that the official test period is over but I'd like to report a debugging bug, how do I do that? I've logged into Premier Support but can't find any software option that allows me to move to the next page.
I don't know what the customer-side of Premier Support looks like anymore - I thought you could select the product you wanted to report against. I don't see the 2017 beta in your list of registered products, though.
If you want to report a problem here, go ahead.
Well I was signed up for the v17 beta program so that's what these bugs are about. You should be able to extract the files in the attached archive and use make to build. Comments in the code describe the two problems:
! [BUG-1]: The following line causes the program to crash !! write(output_unit,fmt='(2(a,dt),a,f6.2)') 'person = ',person,'; abbr = ',abbr,'; score = ',score ! [BUG-2]: with the above line commented out (so that there is no crash) use gdb-ia with a breakpoint on the ! line below. If you examine the contents of person and abbr they are reported as empty whereas the output from ! line 262 shows that they're not empty.
The "crash" is due to a format-variable mismatch error that is raised because you have -check format enabled. That shouldn't happen, though - it's the real(8) SCORE and an F6.2 format, so that should be ok. We'll have to dig into that more.
Still looking at the other issue. I did an initial check on Windows and it's fine, so will have to try gdb-ia next.
The format/variable mismatch happens even without -check format. A workaround is to make the WRITE statement:
write(output_unit,fmt='(a,dt,a,dt,a,f6.2)') 'person = ',person,'; abbr = ',abbr ,'; score = ',score
In other words, don't use the repeat group. I have escalated this as issue DPD200413161. I'll also investigate the debugging problem.