View mode: basic / threaded / horizontal-split · Log in · Help
June 04, 2009
static if syntax
If I write:

static if (cond1)
{
...
}
else static if (cond2)
{
...
}
else
{
...
}

is the block after the final else 'static'? Would it be better if the 'static' before the whole sequence of tests applied throughout rather than having to be explicitly restated in some places, but not in others?

Steve
June 04, 2009
Re: static if syntax
Steve Teale:

>is the block after the final else 'static'?<

Yes, it is.


>Would it be better if the 'static' before the whole sequence of tests applied throughout rather than having to be explicitly restated in some places, but not in others?<

Few times I have forgotten to put a "static" in such chain of static ifs (so I may appreciate a kind of static switch) but in practice I think your idea will lead to other bugs. So it may be better to not change the current design.

Bye,
bearophile
June 04, 2009
Re: static if syntax
Reply to Steve,

> If I write:
> 
> static if (cond1)
> {
> ...
> }
> else static if (cond2)
> {
> ...
> }
> else
> {
> ...
> }
> is the block after the final else 'static'? Would it be better if the
> 'static' before the whole sequence of tests applied throughout rather
> than having to be explicitly restated in some places, but not in
> others?
> 
> Steve
> 

Whatever is used, both of these (and other variations) should be cleanly 
and consistently representable.

static if(foo) {     static if(bar) {} else {}   }
else {    static if(baz) {} else {}   }

static if(foo) {     if(bar) {} else {}   } 
else {     if(baz) {} else {}   }
June 04, 2009
Re: static if syntax
Steve Teale wrote:
> If I write:
> 
> static if (cond1)
> {
>  ...
> }
> else static if (cond2)
> {
>  ...
> }
> else
> {
>  ...
> }
> 
> is the block after the final else 'static'? Would it be better if the 'static' before the whole sequence of tests applied throughout rather than having to be explicitly restated in some places, but not in others?
> 
> Steve

I don't think so.  The problem is that there's no such thing as "chained
ifs" in the language.  What you're actually looking at is this:

static if (cond1)
{
...
}
else
{
static if (cond2)
{
 ...
}
else
{
 ...
}
}

Once you realise that there's nothing special about either "else if" or
"else static if", it makes perfect sense.
June 05, 2009
Re: static if syntax
Daniel Keep Wrote:

> 
> 
> Steve Teale wrote:
> > If I write:
> > 
> > static if (cond1)
> > {
> >  ...
> > }
> > else static if (cond2)
> > {
> >  ...
> > }
> > else
> > {
> >  ...
> > }
> > 
> > is the block after the final else 'static'? Would it be better if the 'static' before the whole sequence of tests applied throughout rather than having to be explicitly restated in some places, but not in others?
> > 
> > Steve
> 
> I don't think so.  The problem is that there's no such thing as "chained
> ifs" in the language.  What you're actually looking at is this:
> 
> static if (cond1)
> {
>  ...
> }
> else
> {
>  static if (cond2)
>  {
>   ...
>  }
>  else
>  {
>   ...
>  }
> }
> 
> Once you realise that there's nothing special about either "else if" or
> "else static if", it makes perfect sense.

Daniel,

So I have to write

static if (cond1)
{
}
else static if (cond2)
{
}
else static if (true)
{
  // for the default alternative?
}

Looks a bit strange
Steve
June 05, 2009
Re: static if syntax
Steve Teale wrote:
> ...
> 
> Daniel,
> 
> So I have to write
> 
> static if (cond1)
> {
> }
> else static if (cond2)
> {
> }
> else static if (true)
> {
>    // for the default alternative?
> }

No.

> Looks a bit strange
> Steve

That's because it's wrong.

static if (cond) ... else ... is a single structure.  The branch is
chosen at compile-time.  It doesn't make sense for the else to somehow
not be part of the static if.
June 05, 2009
Re: static if syntax
Daniel Keep Wrote:


> That's because it's wrong.
> 
> static if (cond) ... else ... is a single structure.  The branch is
> chosen at compile-time.  It doesn't make sense for the else to somehow
> not be part of the static if.

Daniel,

Sorry, I'm there now.

I agree completely that otherwise it does not make sense, but what you say now was not the impression I got from the previous answers.

I had unfortunately, been looking for the appropriate documentation in all the wrong places. A search in the html files comes up with version.html,
which did not seem to be that relevant. Perhaps the file name should reflect the html title, and be conditional-compilation.html.

Thanks Steve
Top | Discussion index | About this forum | D home