Re: [PATCH] Guard stack size warnings with DEBUGMEM

[Jeroen Demeyer <jdemeyer@cage.ugent.be>]

On 2016-01-24 23:49, Bill Allombert wrote:
> However this breaks backward compatibility, since users could
> have set it to 0 in their .gprc, etc.
Sure, but I don't think that backwards compatibility for warnings is so 
important. Besides, what makes you think that users *want* to see those 
warnings anyway?

> This patch is activating by default some warnings that used to be
> only activated when DEBUGMEM>0.
That is trivially solved by changing the debug levels in the code: we 
could change all code of the form
"if (DEBUGMEM > 0)" to "if (DEBUGMEM > 1)" or more generally change
"if (DEBUGMEM > i)" to "if (DEBUGMEM > i+1)".

> In any case I do not think it makes sense to change the interface each
> time someone would like to silence a warning.
But this is a "warning" which can occur when nothing special happens. A 
warning is supposed to actually *warn* about something. This "stack size 
changed to..." is more like an informative message and not a warning.

> We need a global solution for this, preferably with a better
> granularity.
Perhaps it makes sense to do this *in addition*. Even if you add an 
option to silence all warnings, it would still make sense to not always 
show the stack size warnings. I want to see warnings when I'm doing 
something stupid or dangerous, but I don't want to see a "warning" just 
because I changed the stack size.

Jeroen.