[Patch, fortran] PR22304 - gfortran silently changes values in equivalenced variables - MkII

Paul Thomas paulthomas2@wanadoo.fr
Wed Aug 17 20:44:00 GMT 2005


Thomas,

>Your test case doesn't fail for me on i686-pc-linux-gnu without your
>patch (the test case from the PR does).  I would suggest that you
>use the PR testcase, plus additional tests with three or more
>variables.
>  
>
You are quite right.  This test case evolved over time such that, rather 
foolishly, I never checked if the original bug was there or not.
I have to confess that there is something that I do not understand 
here.  I was going to resubmit this, in the next few minutes, as one of 
a series of matches to modules, equivalences and commons.  Before doing 
so, I will understand the story and will resubmit, as a matter of 
course.  According to my understanding, the testcase that I provided 
should have had a double problem... so much for theory!

>This is OK otherwise if regression-tested (unless one of the more
>experienced reviewers has something to say pretty quickly :-)
>
>:REVIEWMAIL:
>  
>
Thanks. As I say above, I will resubmit when I understand what is going 
on and am confident that all is well; I have to say that the patch does 
work with everything that I have devised.

That's a great help - thanks

Paul



More information about the Gcc-patches mailing list