Re: [Gm2] 1. Result from latest tarballsl (30/9/2014 &1/10/2014) for 4.7.3, on i686-32b, under Ubuntu 14.04

[Gaius Mulley]

Gaius Mulley <address@hidden> writes:

> JD <address@hidden> writes:
>
>> Gaius,
>> The summary was:
>>         === gm2 Summary ===
>>
>>     # of expected passes        7561
>>     # of unexpected failures    1513
>>     # of unresolved testcases    540
>>
>> In my build-log the final testsuite lines were:
>>
>>     FAIL: pass/varient5.mod execution, -g 
>>     FAIL: pass/varient5.mod execution, -O 
>>     FAIL: pass/varient5.mod execution, -O -g 
>>     FAIL: pass/varient5.mod execution, -Os 
>>     FAIL: pass/varient5.mod execution, -O3 -fomit-frame-pointer 
>>     FAIL: pass/varient5.mod execution, -O3 -fomit-frame-pointer -
>>     finline-functions 
>>     Running
>>     
>> /home/john/GM2/scratch-area/gcc-4.7.3+gm2-git-latest/gm2/gcc-versionno/gcc/testsuite/gm2/ulmlib/pass/gm2.
>>     exp ...
>>     Running
>>     
>> /home/john/GM2/scratch-area/gcc-4.7.3+gm2-git-latest/gm2/gcc-versionno/gcc/testsuite/gm2/ulmlib/std/pass/gm2.
>>     exp ...
>> So I looked at varient5.mod and tried compiling it - I succeeded! And
>> the a.out "executed" without error - it did bugger all, of course. So,
>> is the testsuite conclusion of "FAIL" wrong?
>>
>> To save you looking, varient5.mod is:
>>
>>     MODULE varient5 ;
>>     
>>     TYPE
>>     Structform = (sets) ;
>>     Structrec = RECORD
>>     CASE form: Structform OF
>>     sets: basep: CARDINAL ;
>>     hmm: RECORD
>>     CASE foo: Structform OF
>>     sets: offset: CARDINAL ;
>>     END
>>     END
>>     END
>>     END ;
>>     
>>     BEGIN
>>     END varient5.
>>     
>> In order to compile it I used gm2 in ~/opt/bin but I needed:
>>
>>     PATH=$PATH:/home/john/opt/lib/gcc/i686-pc-linux-gnu/4.7.3
>> in order to find cc1plus.
>>
>> Is this helpful?
>>
>> Regards,
>> John
>
> Hi John,
>
> yes very useful - thanks!!  I appears that the path set in gm2.exp in
> that directory is incorrect - will fix and commit,

Hi John,

I'm rewriting the gm2_init functions in the testsuite - which is
currently 75% complete - will finish, commit and push changes tomorrow.
Hopefully you should see more regression tests pass after that.  Even in
the worst case - the rationalization in the new gm2_init function will
make it much easier to fix library paths for 100s or 1000s of test cases

regards,
Gaius