I have attached a patch for the problem 192.  The problem is that under mi,
the frame level is being output as an integer followed by a space.  This
occurs because the code in print_frame_info_base was using a format string
%-2d to output the level.  This format has a field-width of 2 and left-adjusts
the output.  This causes a right-most blank to occur for levels < 10.  The
mi code does not analyze format strings such as this and ends up with an
extraneous blank.

To fix this, I have added a new ui_out function: ui_out_field_fmt_int.  It
is essentially the same as ui_out_field_int, but allows specification of
field width and alignment.  The calls to printf_filtered and ui_out_field_fmt
in print_frame_info_base have been replaced with the new function call.
The mi testsuite has been modified to remove expected blanks after frame level.

Is this ok?  Can it be committed?

gdb/ChangeLog:

2002-09-30  Jeff Johnston  <jjohnstn@redhat.com>

	* ui-out.h (ui_out_field_fmt_int): New prototype.
	* ui-out.c (ui_out_field_fmt_int): New function allowing specification
	of field width and alignment.
	* stack.c (print_frame_info_base): When printing frame level, use
	ui_out_field_fmt_int with a width of 2 and left alignment.  Fix for
	PR gdb/192


gdb/testsuite/gdb.mi/ChangeLog:

2002-09-30  Jeff Johnston  <jjohnstn@redhat.com>

	* mi-stack.exp: Change testcases that print frame to not expect "level" field
	to have extraneous right space.  Fix for PR gdb/192.
	* mi-return.exp: Ditto.

-- Jeff J.