Issue 10955 - getopt should have a constraint on the argument count.

[ghost]

http://d.puremagic.com/issues/show_bug.cgi?id=10955

[monarchdodra]

The problem is actually one of passing an odd number of arguments. This currently works:

    auto args = [""];
    getopt(args);

But this fails:

    auto args = [""];
    int i;
    getopt(args, "i", &i, "i");

I think this requires more of an assertive fix that:

1. The length of T is even.
2. The odd arguments are string types.
3. The even arguments are pointer types, functions or delegates.

It's OK if it doesn't compile, as long as the error from a verbose assert.

[ghost]

Well in that case I don't think either approach should be made, since we could add a new backward-compatible API where the oddness/even-ness of the arguments doesn't matter, e.g.:

getopt(args, "count", &count, @doc("this is the thread count"), "name", &name);

This will have 5 arguments. Anyway, depending on particular argument layout with complex rules is not going to make a good API, so maybe we should not touch this code for now.

[ghost]

I forgot to mention: this is with regards to recent NG discussions about adding an optional helper argument to each getopt option, which we haven't decided yet how we're going to implement. But we shouldn't give ourselves double the work by adding complex checking now and removing it later.

[monarchdodra]

But we shouldn't give ourselves double the work by adding complex checking now and removing it later.

True, but I don't think it's too complicated to check each argument exists and is the correct type as you are using/compiling them.

I mean, the error on line 392:

        else
        {
            // it's an option string
            auto option = to!string(opts[0]);
            auto receiver = opts[1];

Could at the very least simply be:

        else
        {
            // it's an option string
            static assert(isSomeString!(T[0]), "unexpected non-string option.");
            auto option = to!string(opts[0]);
            static assert(opts.length > 1, "missing receiver for last option.");
            auto receiver = opts[1];

Sprinkling static asserts never hurt anyone. The current implementation is just assuming everything is perfectly pass, which gives it the impression it is "choking".

[ghost]

Ah yeah, we could add those. This would also fix http://d.puremagic.com/issues/show_bug.cgi?id=10956.