Patch to fix floating point format strings to match fopen documentation

[Mike Miller]

Attached patch changes the string representations of the floating
point format types recognized by octave.  The intent is for the
following to work:

octave:1> fd = fopen('datafile.dat', 'w');
octave:2> [name, mode, mach] = fopen(fd)
name = datafile.dat
mode = wb
mach = ieee-le

so that the machine representation return values match the strings
listed in the fopen help text.  Also, return values are compatible
with Matlab.

Bonus, the following will now work, too, although it's redundant and
less efficient.

octave:3> fd = fopen('datafile.dat', 'w', native_float_format());

This could break any existing code that depends on specific string
values from native_float_format or octave_config_info, I think that is
as far as it goes though.  The contents of those return values are
currently undocumented.

Before:
octave:1> native_float_format ()
ans = ieee_little_endian
octave:2> octave_config_info ('float_format')
ans = ieee_little_endian

After:
octave:1> native_float_format ()
ans = ieee-le
octave:2> octave_config_info ('float_format')
ans = ieee-le

-- 
mike

octave-aa490c112c88.patch
Description: Text Data