PSA: If you run -current, beware!

[David Chisnall]

On 5 Feb 2015, at 07:48, Luigi Rizzo <rizzo at iet.unipi.it> wrote:
> 
> Rather than depending on a compiler option, wouldn't it be better/more
> robust to change ticks to unsigned, which has specified wrapping behavior?

Especially if we want to extend support for external toolchains.  gcc and clang support -fwrapv (though occasionally versions of both will not fully support it), but other compilers may well not have an equivalent.

Translating the code into C is a far more robust solution than the band-aid of telling the compiler to accept a language that is a bit like C and hoping that this will keep working across compiler implementations and versions.

Adding -fwrapv also defeats a number of compiler optimisations, so we are going to generate worse code for places where people used signed types correctly to work around places where they were used incorrectly.

David