RFA: gdb.base/charset.exp: tighten regexp for ^C

[Jim Blandy <jimb@redhat.com>]

The problem with the regexp for the clause that sends ^C is that it
matches a prefix of the correct output.  If expect sees GDB's output
arrive in small chunks, then that clause can trigger even when GDB is
behaving correctly.  This happens on AIX.

I didn't see any discussion in the archives of the sort of situations
this clause was meant to handle.  A different approach might be for
the timeout clause to send a ^C the first time around, and then
actually fail the second time around.  Then the situation the clause
the patch tweaks was meant to handle (now there's a nice noun clause)
will be caught by the ordinary $gdb_prompt case.

2004-06-03  Jim Blandy  <jimb@redhat.com>

	* gdb.base/charset.exp: Only send a control-C if we see a new
	prompt and incomplete command.

Index: gdb/testsuite/gdb.base/charset.exp
===================================================================
RCS file: /cvs/src/src/gdb/testsuite/gdb.base/charset.exp,v
retrieving revision 1.3
diff -c -p -r1.3 charset.exp
*** gdb/testsuite/gdb.base/charset.exp	26 Feb 2004 17:23:23 -0000	1.3
--- gdb/testsuite/gdb.base/charset.exp	4 Jun 2004 04:20:01 -0000
*************** gdb_expect {
*** 140,146 ****
  	# We got some output that ended with a regular prompt
          fail "get valid character sets"
      }
!     -re "^set charset.*$" {
  	# We got some other output, send a cntrl-c to gdb to get us back
          # to the prompt.
  	send_gdb "\003"
--- 140,146 ----
  	# We got some output that ended with a regular prompt
          fail "get valid character sets"
      }
!     -re ".*$gdb_prompt set charset.*$" {
  	# We got some other output, send a cntrl-c to gdb to get us back
          # to the prompt.
  	send_gdb "\003"