[Ns-developers] 64 bit port of ns

[Mathieu Lacage]

On Thu, 2005-09-01 at 08:12 +0100, Lloyd Wood wrote:
> > That right. This did not make any sense (the method is overloaded by
> > subclasses but is not overriden. I think the author was confused about
> > these two different concepts) so I removed the virtual from the method
> > declaration. I probably should refrain from making such unrelated
> > changes in my patches.
> 
> That would be a good idea - this probably breaks third-party changes

Unfortunately, my change already breaks third-party code which tries to
override this method in very un-obvious and hard-to-detect ways. There
is nothing to do about it unless you don't want to fix these 64 bits
issues or unless you want to add even more compatibility crap in this
header file.

> to ns that do override the method. 'Unnecessary' virtual declarations
> allow ns to be extended more easily.

Well, unless you know of a real user or you know how this keyword came
to be here (I have no access to the cvs history, so I cannot even go do
a bit of archeology), and since there is no comment explaining why it is
here, trying to guess whether or not it is used out-of-tree is pretty
much a useless exercise. As such, I will refrain from including here a
detailed analysis of the psychology of the author of that piece of code
based on the code itself :)

In that context, it was clearly not a great idea to do what I did but I
believe that this kind of "fix" should be attempted over the whole
codebase in an attempt to cleanup at least a bit of cruft. If users
complain about some breakage, we can easily find a better fix for them. 

As-is, this classifier code looks like a bad joke and I am getting sick
of spending my days reading crap like that (or worse) without being able
to do anything about it. Yes, in case you have noticed, I am slightly
pissed off (not at you or anyone in particular, just at my waste of
time).

Here is a "fixed" patch which removes the offending part of the original
patch. Thanks to pavel for catching this issue.

http://www-sop.inria.fr/dream/personnel/Mathieu.Lacage/ns2/64-1.patch

regards,
Mathieu
--