Re: [rfc] clean up linespec.c

[Elena Zannoni <ezannoni at redhat dot com>]

Daniel Jacobowitz writes:
 > On Thu, Nov 07, 2002 at 12:30:26PM -0500, Elena Zannoni wrote:
 > >  > It turns out that decode_line_1 isn't quite as crazy a function as I'd
 > >  > feared.  There's a reasonable flow of control underneath (well, at
 > >  > least there is once you get rid of the unnecessary goto's), though
 > >  > admittedly the C++ part of the function is still pretty complicated,
 > >  > and the function will always consist of a bunch of special cases.
 > >  > 
 > > 
 > > Having c++ separated in functions, is a first step in moving C++
 > > support to its own files.....:-)
 > 
 > :)
 > 
 > >  > I hope I didn't break anything, though my only real evidence for that
 > >  > is that I didn't get any new regressions on the testsuite.  (I have no
 > >  > idea how comprehensive the testsuite's coverage of linespec is.)
 > > 
 > > ah, gcov data would be useful here... :-)
 > 
 > Gcov data is good and all, but it's a required-but-not-sufficient for
 > coverage testing.  I've been doing some coverage tests on
 > c-typeprint.c, and I would never have found the char *constvarname bug
 > that was fixed recently.
 > 

Yeah, also things get complicated because gdb has so many different
targets that trigger so many different code paths, that you are never
sure if some code is really never covered.

 > It's still impressively useful though :)
 > 

true.

Elena


 > 
 > -- 
 > Daniel Jacobowitz
 > MontaVista Software                         Debian GNU/Linux Developer