Re: [PATCH/RFA] PR gdb/648 (eval.c approval reqd)

[Andrew Cagney <cagney@gnu.org>]

>>> It seems to me that there's a separate bug in f-valprint.c.  The
>>> output below is incorrect, right?  It's printed as a series of
>>> columns, not a series of rows.  Or is that what users of a Fortran
>>> debugger would expect?
>>> 
>>>     (gdb) p twodi
>>>     $1 = (( 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18,
>>>     19, 20) ( 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30, 32,
>>>     34, 36, 38, 40) ( 3, 6, 9, 12, 15, 18, 21, 24, 27, 30, 33, 36, 39, 42,
>>> [..]
> 
> 
> A Fortran user would expect this like this, so it's not a bug.  Suppose
> a Fortran-er is working with an array of 3-d points, for some finite
> element application.  To ensure the x,y,z are in adjacent memory for
> better cache behaviour, there'd be an array of REAL(3, 100000) -- which
> when printed by GDB will produce a list of points (x, y, z). They'd want
> that.

Where were we at with the testcase?  I've a fuzzy memory of it hitting a 
problem with test infrastructure?

Andrew