Notes from C++ static analysis

[Adam D. Ruppe]

On Wednesday, 26 June 2013 at 18:54:17 UTC, H. S. Teoh wrote:
>> import std.stdio;
>> void main() {
>>     size_t x = 10;
>>     writefln("%d", x, x);
>> }
>> 
>> In a modern statically typed language I'd like such code to 
>> give a compile-time error.
>
> This looks like a bug to me. Please file one. :)

Not necessarily, since you might want a format string to be a 
runtime variable, like when doing translations. I could live with 
there being another function that does runtime though.

Things might be confusing too because of positional parameters 
(%1$d). You might offer something that isn't necessarily used:

config.dateFormat = "%3$d/%2$d";
writeln(config.dateFormat, year, month, day);

> Anyway, in the end I decided that naming member function 
> arguments after member variables is a Very Stupid Idea,

Blargh, I do it a lot. But I would be ok with the lhs of a member 
when there's a parameter of the same name requiring that you call 
it this.x.

> How would the compiler catch such cases in general, though? I 
> mean, like in arbitrarily complex boolean expressions.

The Microsoft compiler warned about it, after constant folding, 
working out to if(1).

I'm a little concerned that it would complain about some false 
positives though, which can be quite deliberate in D, like 
if(__ctfe).