Silverfrost Forums

Welcome to our forums

Error 1017

4 Jul 2021 4:36 #28051

When I try recompiling an old project to create a .dll for .NET, I get the following error for few subroutines:

'error 1017: Cannot yet process records of type force_temporary in CLR mode'

Line reference is to the sub declaration.

What does this error refer to?

5 Jul 2021 5:52 #28052

Emanuele

This is an internal compiler error. We would need some Fortran code that generates this error in order to comment further.

7 Jul 2021 9:54 #28056

Thanks Paul, I am trying to isolate the problem, since the code is composed by many routines...

7 Jul 2021 9:36 #28057

I managed to reproduce the error with these lines of code:

subroutine test (avis)
real avis, alti, re, gam

alti=35785.845
re=6378.155
gam=0.1
avis=asin((1.+alti/re)*(gam))

return
end

I'm working on VS2019 with FTN95 8.70 PE. Compiler switches as follows: /DEBUG /CHECK /FULL_DEBUG /WIDE_SOURCE /STANDARD_ERROR_REPORTS /SINGLE_THREADED /FPP /CLR /CLR_VER 4

The problem seems to be related with the asin() function.

Thanks for your help.

8 Jul 2021 6:42 #28058

Emanuele:

The bug is also present in Version 8.75. Here is a work-around for now.

subroutine test (avis)
real, intent(out) :: avis
real alti, re, gam, arg

alti=35785.845e0
re=6378.155
gam=0.1
arg = (1.+alti/re)*(gam)
avis=asin(arg)

return
end
8 Jul 2021 6:45 #28059

Emanuele

The quick fix for this failure is to remove the brackets from (gam) on line 7...

avis=asin((1.+alti/re)*gam)
8 Jul 2021 10:21 #28061

Thank you very much guys! I never would have guessed it!

I am a bit scared because potentially there can be dozens of such situations in the original code! :lol:

A curiosity: is it possible that in older versions this bug was not present?

Thanks again.

8 Jul 2021 11:57 #28062

Emanuele

Yes it is quite possible that this would not have failed with earlier versions.

8 Jul 2021 1:15 #28065

Quoted from Emanuele A curiosity: is it possible that in older versions this bug was not present?

The error does not occur with FTN95 7.2 (32-bit only). This finding agrees with what Paul said.

When the bug bites, you get an error message, don't you? Therefore, there is no reason to worry, unless you see some indication that an apparently successful compilation has produced bad code.

Please login to reply.