Re: [RFA] make first parameter of to_lookup_symbol const char *

[Stan Shebs <stan at codesourcery dot com>]

On 3/16/11 6:58 AM, Joel Brobecker wrote:
> > > BTW, it looks like no target defines this operation...
> > Unless you use it for some new patch of yours it should be removed instead.
> This is what it looks like to remove the target_ops method. It feels
> a little like excising a potentially useful feature, so I'm not going
> to commit without review, although there is no sign that we'll ever
> need it any time soon.  But I added a comment explaining what we used
> to do, to give us a clue later on, if we encounter a target where
> we might need something of this kind.

Yeah, it looks like the last use quietly evaporated with the deletion of 
remote-vx.c in 2004.

Features without any means of exercise are likely to bitrot semantically 
even if they continue to compile/run (witness tracepoints), so it's in 
our interest to be ruthless.

I used to have fun using gcov with the GDB testsuite to find segments of 
code that were never exercised - gcov results are cumulative, so the 
post-testing coverage display is good for ideas as to what test cases 
ought to be written, and also suggests code that might turn out to be 
intrinsically unreachable.

Perhaps msnyder can play with that when he gets tired of coverity. :-)

Stan