Re: [Vala] R: Re: R: POSIX and Dova profiles

[Jürg Billeter <j bitron ch>]

On Fri, 2012-08-03 at 17:42 +0200, Andrea Del Signore wrote:
> On Fri, Aug 3, 2012 at 5:26 PM, Jürg Billeter <j bitron ch> wrote:
>         Unused parts of the library will not have any impact on the
>         size of the binary in the current plan.
>
> How will this be achieved? Dead function elimination by the C compiler
> and multiple static libraries like in Dova?

At least one possibility will be to directly build from library sources
and application sources, in which case dead code will not even generate
C code. This will not only eliminate unused classes but also unused
methods of used classes. In case you're wondering, the base library will
have a permissive license (MIT/BSD).

>         Are you referring to a C backend or to C-like syntax?
>
> C-like syntax ;) , but I think that the C backend will remain right?

It will still use braces but the syntax will be closer to Go in some
aspects. The initial release is expected to contain a C backend, which I
expect to stay around. However, the compiler is designed to make it
relatively easy to add other backends for LLVM IR or even machine code.

>         > (On a side note I'm not sure that I like the result = blah of the dova
>         > profile, it reminds me vb6 when you have to write funcion_name = blah,
>         > and I think that return blah is still better and more clear)
>         
>         The current plan is to move back to return statements, although possibly
>         with a caveat that the return will not be allowed to act as an arbitrary
>         jump.
>  
> Cool, what will substitute the arbitrary jump then?

The usual control flow statements such as 'if' and 'switch'. The return
statement will simply have to be at a point where the succeeding block
is already the exit block.

Jürg