I'm looking at setting priorities for finishing up a beta compiler.

What features do people need the most to be implemented for a credible beta?

-Walter

"Walter" <walter@digitalmars.com> wrote in message news:a71k32$2lne$1@digitaldaemon.com...

> I'm looking at setting priorities for finishing up a beta compiler.
>
> What features do people need the most to be implemented for a credible
beta?

I'd personally like to see property settors & gettors, and something better than printf & scanf for console I/O.

I'd like to see associative arrays flushed out for types other than int[char[]].


"Walter" <walter@digitalmars.com> wrote in message news:a71k32$2lne$1@digitaldaemon.com...
> I'm looking at setting priorities for finishing up a beta compiler.
>
> What features do people need the most to be implemented for a credible
beta?
>
> -Walter
>
>

"Walter" <walter@digitalmars.com> wrote in news:a71k32$2lne$1@digitaldaemon.com:

> I'm looking at setting priorities for finishing up a beta compiler.
> 
> What features do people need the most to be implemented for a credible beta?
> 
> -Walter

More important than features, I think is licensing.
The compiler will be freeware ? Open Source ? You wanna make a business around the compiler ? How can this benefit a
comunity and your work ?

Walter wrote:
> I'm looking at setting priorities for finishing up a beta compiler.
> 
> What features do people need the most to be implemented for a credible beta?
> 
> -Walter

I know this is something that nobody likes to do, but the documentation could probably use a going-over.  Seeing as how this is a new language that nobody is familiar with - you'd really need good docs to find out what the features are so you could try them, but not be misled by references to things that were planned but didn't make the final cut.

For example: on the "attributes" page "Static, Final, Abstract Attributes" are just described with "To be written".  I tried declaring something static, assuming that it would have the same meaning as in C regarding linkage, but it didn't.  Was that because "static" means something slightly different in D? or was it a bug/unimplemented feature?  Without the docs I had no way to really tell.

The "classes" page mentions "D classes support the single inheritance paradigm, extended by adding support for interfaces", but doesn't really say much more about interfaces.  As somebody with a Java background, I'd want to explore that, but there's nothing else to go on.

Things like the bit on " Strings can be ... compared ... if (str1 < str3)" which was discussed in a different thread need to looked at depending on if that feature really is there or not.

The page on "interfacing to C" is mostly incomplete, I think especially the bit about calling printf with D strings using the "%.*s" specfier really needs to be repeated there (or fixed) since that's going to burn a lot of people taking a first look at D.

The page on the DMD alpha says " Run: \dmd\bin\shell all.sh in the \dmd\samples\d directory for several small examples.", but that doesn't actually work out-of-the-can - which is a bit vexing to somebody first taking a look at this software.

1) The comment in "all.sh" about the path to shell.exe needs one more "..". The declaration of "MARS" probably should have "-I\dmd\src\phobos" added, and some mention should be made of also needing to set LIB to \dmd\lib so "sc" can find "phobos.lib", and adding "\dm\bin" to the path so the script can find "sc" (or fixed with declarations in the script).

2) And if you do make those fixes, the script still bails out trying to compile hello2.html with the error "\dmd\src\phobos\object.d(2): symbol printf symbol object.printf conflicts with Object.printf at \dmd\src\phobos\Object.d(2)"

3) The chello.d sample fails since there's no "dserver.def" file

-----

Anyhow, those are a few things have had me scratching my head a bit, and might cause other people problems too.

	Barry

"Patrick Down" <pdown@austin.rr.com> wrote in message news:a72inq$5bu$1@digitaldaemon.com...

> I'd like to see associative arrays flushed out for types other than int[char[]].

Oh yes, forgot to mention this!

"Juarez Rudsatz" <juarez@correio.com> wrote in message news:Xns91D490A5A8A1Djuarezcom@63.105.9.61...

> More important than features, I think is licensing.
> The compiler will be freeware ? Open Source ? You wanna make a business
around the compiler ? How can this benefit a
> comunity and your work ?

I don't know for sure, but I guess the licensing schema will be the same as for Digital Mars C++ - you get the compiler for free, and for the IDDE, you're gonna pay (once Walter adds D support for it).

Oh, and Dfront will be open-sourced, or so it seems from posts on the topic?

"Barry Pederson" <barryp@yahoo.com> wrote in message news:3C94E64F.3090702@yahoo.com...

> I know this is something that nobody likes to do, but the documentation
could
> probably use a going-over.  Seeing as how this is a new language that
nobody
> is familiar with - you'd really need good docs to find out what the
features
> are so you could try them, but not be misled by references to things that
were
> planned but didn't make the final cut.

Oh yes. There is quite a lot of things in the docs that were previously correct, in the earlier versions, but now are gone...

> For example: on the "attributes" page "Static, Final, Abstract Attributes"
are
> just described with "To be written".  I tried declaring something static, assuming that it would have the same meaning as in C regarding linkage,
but it
> didn't.  Was that because "static" means something slightly different in
D? or
> was it a bug/unimplemented feature?  Without the docs I had no way to
really tell.

"static" in D is for locals and class members only, and its meaning is the same as in C++. To hide globals, you use "private".

Oops, forgot about static constructors & destructors... =)

> The "classes" page mentions "D classes support the single inheritance paradigm, extended by adding support for interfaces", but doesn't really
say
> much more about interfaces.  As somebody with a Java background, I'd want
to
> explore that, but there's nothing else to go on.

There isn't much to explore, in fact. The interface is just that - an interface, with a set of methods:

    interface IFoo
    {
        int bar(int);
        void baz(char[]);
    }

Usage of interfaces is also trivial:

    class Foo: Object, IFoo
    {
        int bar(int n) { ... }
        void baz(char[] s) { ... }
    }

Not much different from Java...

On Sun, 17 Mar 2002 03:21:28 -0500, Walter wrote:

> I'm looking at setting priorities for finishing up a beta compiler.
> 
> What features do people need the most to be implemented for a credible beta?
> 
> -Walter

source code and linux support. :-)

Pavel Minayev wrote:

> There isn't much to explore, in fact. The interface is just that - an
> interface, with a set of methods:
> 
>     interface IFoo
>     {
>         int bar(int);
>         void baz(char[]);
>     }
> 
> Usage of interfaces is also trivial:
> 
>     class Foo: Object, IFoo
>     {
>         int bar(int n) { ... }
>         void baz(char[] s) { ... }
>     }
> 
> Not much different from Java...

OK, that's good to know, but my point was that the docs are lacking even that little bit of info or simple example declaration.  If somebody's evaluating this language, they're gonna want to see that documented a bit, rather than just guessing how it should work, or searching through newsgroup lore.

Also, the page on syntax grammar doesn't contain any definition for interface declarations (although the lexical grammar page does mention "interface" as a keyword).

	Barry