|
From: Greg P. <gp...@ap...> - 2009-08-24 22:43:14
|
On Aug 24, 2009, at 7:43 AM, Julian Seward wrote: > On Sunday 23 August 2009 12:16:33 am Nicholas Nethercote wrote: >> On Sat, Aug 22, 2009 at 8:33 PM, Bart Van Assche wrote: >>> I agree that most if not all VG_(printf)() calls should be >>> replaced by >>> VG_(message)() calls, but I'm not sure that it is a good idea to >>> rename VG_(printf)() into something else. >> >> The motivation is this: if it should only be used in rare cases, >> then >> give it a non-standard name so that people don't use it by default >> without thinking. > > Overall I'm in favour of the changes you suggest (making less use of > VG_(printf)). But I'd prefer not to rename it to > VG_(something_so_totally_obscure_youd_never_think_to_use_it). How > about > VG_(dprintf), the d to emphasise it's debuggingness? I think > dprintfs are > a fairly widely understood thing. If it's intended for development use only, it could be marked __attribute__((deprecated)), which would in turn cause a build warning everywhere it's called. When approaching a release, make sure there aren't any of those warnings left. -- Greg Parker gp...@ap... Runtime Wrangler |