Re: [lwip-users] RFC: Debug changes

[David Haas]

I can certainly organize the variables in an lwip_debug structure to 
avoid the name collision issue. We can still have a #define LWIP_DEBUG 
which would turn on/off the whole thing. So if LWIP_DEBUG is not 
defined, there will be no debugging. If LWIP_DEBUG is defined though, 
you will use the lwip_debug.mem_debug (or similar) variable to control 
if output occurs for that specific area.

MEM_DEBUG and similar macros can control whether lwip_debug.mem_debug 
defaults to 1 or 0.

David.



Marc Boucher wrote:

> since everything generally gets linked together
> on most platforms without namespaces, generic names
> like mem_debug should be avoided.
>
> Perhaps all lwip debug variables could be grouped into
> an array or structure (lwip_debug) to avoid name collisions
> and make it easy to find/change them in a central location
> with debuggers etc..
>
> Also perhaps the macros could be organized in a way allowing
> those who do not want/need runtime debug control to
> override them so as to revert to the current behavior.
>
> Marc
>
>
> On Wed, Feb 12, 2003 at 02:32:16PM -0500, David Haas wrote:
>  
>
> > Hi all:
> >
> > Would anyone have any objections if I changed the debug interface to 
> > support variables controlling whether debug is output or not. In words, 
> > instead of having a #define MEM_DEBUG, we would have a mem_debug global 
> > variable which could be TRUE or FALSE. In order to enable debug output 
> > you would then just set mem_debug to TRUE. This could be done 
> > dynamically without rebuilding.
> >
> > All of this functionality would be built into a new debug.c file.
> >
> > David.
> >
> >
> >
> >
> > _______________________________________________
> > lwip-users mailing list
> > address@hidden
> > http://mail.nongnu.org/mailman/listinfo/lwip-users
> >
> >    
>
>
>