[wx-dev] Remove meaningless assert in wxDebugReport/wxStackWalker

[Alex Bligh]

Vadim Zeitlin wrote:
> On Fri, 04 Aug 2006 19:27:29 +0100 Alex Bligh <alex at alex.org.uk> wrote:
> 
> AB> 00400000-00414000 r-xp 00000000 08:01 13533240                           /home/amb/camelot/wxWidgets/WX_2_6_BRANCH/debug/samples/debugrpt/debugrpt
> AB> 00513000-00515000 rw-p 00013000 08:01 13533240                           /home/amb/camelot/wxWidgets/WX_2_6_BRANCH/debug/samples/debugrpt/debugrpt
> AB> 00515000-00931000 rw-p 00515000 00:00 0                                  [heap]
> 
>  Ok, thanks. Unfortunately I'm none the wiser... Anyhow, if this does
> happen, we need to handle this situation and it probably means that we
> should create multiple entries for the same module: it doesn't seem right
> to pretend that the range (00414000, 00513000) belongs to this module when
> it doesn't. Do you agree?

Well, I'm not sure it will create any harm doing that because by it's
very nature, if there's a hole, it shouldn't appear on the stack walk.
However, if we had modules in the order A <hole> B <hole>
A <hole> B <hole> I can see it would definitely be a dubious guess.

I'm presuming the clue here as to why there are two entries is
that one is "r-xp" which is the executable, and one is "rw-p"
which is presumably static init or similar. I didn't do anything
fancy to create this, just compiled with gcc-4.0.3 out the box.

So whilst I think you are right that a multiple entries for
the same module is probably the way to go, I don't really see
a problem just commenting out the wxAssert for now given it
does no good, and a fair amount of harm (see the O(n^2) bug).
Or am I just being lazy? :-)

Alex