gfortran: fatal internal error - report requested

[John Watts]

1. Ordering issue:    Of course - That's why the Makefile variant 
doesn't complain - it knows all about ordering.    But the compiler 
shouldn't have a nervous breakdown about this - it should just politely 
point out the error of its user's ways.
2. No report would have been submitted had it not been requested
3. As for using a pre-release compiler - I only downloaded it a few days 
ago and it looked like the latest thing at that time and URL...!  
4. No ICE?   Good.   I'll try to find a later variant.

Daniel Franke wrote:
> On Monday 21 December 2009 01:26:02 Steve Kargl wrote:
>   
>> On Sun, Dec 20, 2009 at 04:18:55PM -0800, Tim Prince wrote:
>>     
>>> Steve Kargl wrote:
>>>       
>>>> You have an ordering issue.  You can't USE the module
>>>> until it's compile.  Your main program will be compiled
>>>> before the MODULE TEST_MODULE.  Chicken meet egg.
>>>>         
>>> However, the internal error, if it persisted in a current version, would
>>> still be a bug, even though provoked by wrong ordering.  8 months is a
>>> long time to use a pre-release compiler.
>>>       
>> There is no ICE.
>>     
>
> On Sunday 20 December 2009 22:45:53 John Watts wrote:
>   
>>> gfortran -v
>>>       
>> Target: i586-pc-mingw32
>>     
>
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31519
> 	Summary: spurious ICE messages when module does not compile"
> 	Target:  i386-pc-mingw32
>   

-- 

Regards,

John Watts



Everything that is Proveable is True.   
But not everything that is True is Proveable - and we can prove that.