Testsuite failure (mi-var-child-f.exp)

Testsuite failure (mi-var-child-f.exp)

[Nick Roberts]

I get the following fail:

Running /home/nickrob/src/gdb/testsuite/gdb.mi/mi-var-child-f.exp ...
FAIL: gdb.mi/mi-var-child-f.exp: create local variable array
FAIL: gdb.mi/mi-var-child-f.exp: get children of array
FAIL: gdb.mi/mi-var-child-f.exp: get grandchildren of array (children of first element of second index)

		=== gdb Summary ===

# of expected passes		2
# of unexpected failures	3
/home/nickrob/src/gdb/testsuite/../../gdb/gdb version  6.5.50.20061029-cvs -nx


I didn't see this before because when I ran the testsuite it was untested
as array.f couldn't compile.  I have no file called g77 and perhaps others
don't too.  Should runtest also look for gfortran now?

array.f (used by mi-var-child-f.exp) has the line:

      DATA array/11,21,12,22,13,23/


In response to:

  -var-create - * array

I now get (gfortran):

  ^done,name="var1",numchild="6",type="int4 (0:5)"

when I used to get (g77):

^done,name="var1",numchild="3",type="int4 (2,-1:1)"


I'm pretty sure this is a problem with gfortran (incorrect DWARF output
or something) as it also happens with gfortran and an old GDB.

So that I'm not told it's a GDB issue by gfortran developers, can someone
on this list identify exactly what the problem is please?


-- 
Nick                                           http://www.inet.net.nz/~nickrob

Re: Testsuite failure (mi-var-child-f.exp)

[Nick Roberts]

Nick Roberts writes:
 > array.f (used by mi-var-child-f.exp) has the line:
 > 
 >       DATA array/11,21,12,22,13,23/

Sorry, I meant to refer to the line:

      INTEGER array(1:2,-1:1)


-- 
Nick                                           http://www.inet.net.nz/~nickrob

Re: Testsuite failure (mi-var-child-f.exp)

[Daniel Jacobowitz]

On Tue, Oct 31, 2006 at 09:43:58PM +1300, Nick Roberts wrote:
> I'm pretty sure this is a problem with gfortran (incorrect DWARF output
> or something) as it also happens with gfortran and an old GDB.
> 
> So that I'm not told it's a GDB issue by gfortran developers, can someone
> on this list identify exactly what the problem is please?

They won't tell you that; they're aware that gfortran's debug info is
mostly junk.  Steven Bosscher has been working on it.

-- 
Daniel Jacobowitz
CodeSourcery

Re: Testsuite failure (mi-var-child-f.exp)

[Nick Roberts]

 > They won't tell you that; they're aware that gfortran's debug info is
 > mostly junk.  Steven Bosscher has been working on it.

It worked for g77, and I thought gfortran was a fork of that.  Anyway are
you saying there is little point in reporting it since there are many
other problems with the debug info?

Also how about my other point:  Should the testuite find gfortran or is that
a DajaGnu issue?


-- 
Nick                                           http://www.inet.net.nz/~nickrob

Re: Testsuite failure (mi-var-child-f.exp)

[Daniel Jacobowitz]

On Wed, Nov 01, 2006 at 09:24:54AM +1300, Nick Roberts wrote:
>  > They won't tell you that; they're aware that gfortran's debug info is
>  > mostly junk.  Steven Bosscher has been working on it.
> 
> It worked for g77, and I thought gfortran was a fork of that.

No, they're basically unrelated.

> Anyway are you saying there is little point in reporting it since
> there are many other problems with the debug info?

No, the opposite.  There's a meta-bug in the GCC bugzilla tracking
debug info issues.  If there isn't already one for array indexes, just
file a new one.

> Also how about my other point:  Should the testuite find gfortran or is that
> a DajaGnu issue?

I don't know.  It seems like testing gfortran would be more useful than
g77 nowadays.

-- 
Daniel Jacobowitz
CodeSourcery