|
From: Nicholas N. <nj...@cs...> - 2005-03-03 00:38:59
|
Hi, Can someone remind me why we use mangled C++ names in suppressions? I just tried switching to using demangled names and it worked fine, and the suppressions looked much better. Thanks. N |
|
From: Julian S. <js...@ac...> - 2005-03-03 00:49:24
|
On Thursday 03 March 2005 00:38, Nicholas Nethercote wrote: > Hi, > > Can someone remind me why we use mangled C++ names in suppressions? I > just tried switching to using demangled names and it worked fine, and the > suppressions looked much better. Not sure. Perhaps because it might have been considered a possible performance problem? Or more likely because it never got properly thought about before? J |
|
From: Jeremy F. <je...@go...> - 2005-03-03 00:51:40
|
Nicholas Nethercote wrote:
> Can someone remind me why we use mangled C++ names in suppressions? I
> just tried switching to using demangled names and it worked fine, and
> the suppressions looked much better.
Do we have a mangler in there? Or does it run the demangler and then
compare against the un-mangled name?
Uh, maybe we've just been dense; I can't see why this wouldn't work.
J
|
|
From: Nicholas N. <nj...@cs...> - 2005-03-03 01:28:57
|
On Wed, 2 Mar 2005, Jeremy Fitzhardinge wrote: >> Can someone remind me why we use mangled C++ names in suppressions? I >> just tried switching to using demangled names and it worked fine, and >> the suppressions looked much better. > > Do we have a mangler in there? Or does it run the demangler and then > compare against the un-mangled name? > > Uh, maybe we've just been dense; I can't see why this wouldn't work. It's a two line change in vg_errcontext.c -- change the two calls to VG_(get_fnname_nodemangle)() to VG_(get_fnname)(). Of course now it's hard to change, because everyone has their pre-existing suppression files. We could try matching against both the mangled and demangled names, I guess, and for auto-generation start using the demangled names as of now. N |
|
From: Nicholas N. <nj...@cs...> - 2005-03-03 05:05:04
|
On Wed, 2 Mar 2005, Nicholas Nethercote wrote: >>> Can someone remind me why we use mangled C++ names in suppressions? I >>> just tried switching to using demangled names and it worked fine, and >>> the suppressions looked much better. >> >> Do we have a mangler in there? Or does it run the demangler and then >> compare against the un-mangled name? >> >> Uh, maybe we've just been dense; I can't see why this wouldn't work. > > It's a two line change in vg_errcontext.c -- change the two calls to > VG_(get_fnname_nodemangle)() to VG_(get_fnname)(). To answer your question, we would run the demangler more often than now. But I don't think it's a problem. N |