Re: [PATCH v2 3/5] gdb/testsuite: Add gdb.arch/aarch64-mops-watchpoint.exp

[Tom Tromey <tom@tromey.com>]

>>>>> "Tom" == Tom de Vries <tdevries@suse.de> writes:

Tom> - obviously, using a gdb_exit in a proc results in trouble if the proc
Tom>   is called at a point in the test-case where we don't want a gdb_exit
Tom> - using a gdb_exit in a caching proc may or may not result in trouble if
Tom>   the proc is called at a point in the test-case where we don't want
Tom>   gdb_exit.  Whether we run into trouble depends on whether the result
Tom>   of the proc is already cached or not.  This behaviour is undesirable.

I didn't really follow this thread, but I tend to think that caching
procs should be in two categories: those that require gdb to already be
running, and those that are standalone.  Having a proc like this call
gdb_exit seems like a recipe for some confusion, to me.  Even having one
start a new gdb seems like trouble, since IIRC that's hard to do without
introducing new test results, but the nature of a caching proc is that
the test names would then depend on which process wins the race to run
it first.  I've thought sometimes about adding some mechanism to enforce
this split.

Tom