|
From: Tom S. <to...@pl...> - 2006-11-03 20:28:32
|
On Sat 06-11-04 06:56, Nicholas Nethercote wrote: > On Fri, 3 Nov 2006, Tom Schutter wrote: > > > The valgrind-3.2.1 man page says this about --undef-value-errors: > > --undef-value-errors=<yes|no> [default: yes] > > Controls whether memcheck detects dangerous uses of undefined > > value errors. When yes, Memcheck behaves like Addrcheck, a > > lightweight memory-checking tool that used to be part of > > Valgrind, which didn't detect undefined value errors. Use this > > option if you don't like seeing undefined value errors. > > > > I would think that --undef-value-errors=yes would cause memcheck to > > detect undefined value errors, just based upon the name of the option. > > But the man page implies that it is the other way around. > > > > Also, the "Use this option if you don't like..." sentence is > > ambiguous. Does it mean use "--undef-value-errors=yes" if you don't > > like, or use "--undef-value-errors=no" if you don't like? > > > > Does selecting yes or no affect the memcheck runtime? If so, maybe > > the doc/manpage should mention that as well. > > > > I would also suggest dropping the reference to Addrcheck, it doesn't > > help new users at all. > > You're right, the description is backwards and confusing. Here's the new > one I just committed: > > <para>Controls whether <constant>memcheck</constant> detects > dangerous uses of undefined value errors. Set this to > <varname>no</varname> if you don't like seeing undefined value > errors; it also has the side effect of speeding > <constant>memcheck</constant> up somewhat. > > Nick Note that it also appears in the doc as well, see http://www.valgrind.org/docs/manual/mc-manual.html#mc-manual.flags -- Tom Schutter (mailto:to...@pl...) Platte River Associates, Inc. (http://www.platte.com) |