Hello, all.

    First of all, sorry my english. But I thing it's enough to keep a
communicate with you.
    I've found the D language at google last week while I was searching for
some programming stuff.
    So after reading some of its features I became very interested because
it really brings to the programmer a lot of good things. I've read all the
documentation and I'm interested on spending some of my time on the compiler
development.
    But for doing that, I need some support from you.
        1) I have a good programming skill, but I haven't work on a compiler
yet, and I don't know some terms: What does mean "build a D frontend for GCC
backend.". The "frontend" would generate some kind of pseudocode for the
"backend" process?
        2) Although having Linux installed on my computer, I mainly work on
Windows XP OS. For doing this job, would I have to use Linux or cygwin? Is
there any "native" gcc port for Win32 (I mean, some gcc working without
cygwin)?
        3) I've seen that the language specification for D is not complete.
When is it planned to be?
    That's all, for now.

    Luciano.

Luciano Coutinho Barcellos wrote:
>         1) I have a good programming skill, but I haven't work on a compiler
> yet, and I don't know some terms: What does mean "build a D frontend for GCC
> backend.". The "frontend" would generate some kind of pseudocode for the
> "backend" process?

Front end is the parser which creates a 'tree' type of structure. The 'tree' is handed to the backend which generates the code.

>         2) Although having Linux installed on my computer, I mainly work on
> Windows XP OS. For doing this job, would I have to use Linux or cygwin? Is
> there any "native" gcc port for Win32 (I mean, some gcc working without
> cygwin)?

I personally prefer a BSD Unix www.freebsd.org, www.openbsd.org, www.netbsd.org, but any stable Linux should do.

>         3) I've seen that the language specification for D is not complete.
> When is it planned to be?

No idea. I don't think Walter has a target date, which is basically a great thing as I think this allows the language to evolve and de-volve where necessary. In other words, it allows to try something and see if it works and if not it can be rolled back or changed.

Thanks!
Jan


-- 
ManiaC++
Jan Knepper

Hi. Thanks for your will to help and welcome aboard!

Luciano Coutinho Barcellos wrote:
>         2) Although having Linux installed on my computer, I mainly work on
> Windows XP OS. For doing this job, would I have to use Linux or cygwin? Is
> there any "native" gcc port for Win32 (I mean, some gcc working without
> cygwin)?

Jan has an exceptional ability to answer wrong questions, which have not even been asked. :)

So, the native GCC for Windows is called MinGW (www.mingw.org). It is recommended that you use it, but you may just as well use DigitalMars C++ compiler in the first stages of development.

The DMD frontend (which is open source) has got no machine interface header. These can be ported from the old DLI compiler (www.opend.org). Then a simple "machine" should be creaeted - that is a compiler back-end - which would not be bound to any real code generator, but simply print what it gets from the front-end. Then we decide what to do next. But until then, you can use any C++ compiler.

I'm looking for a suitable toolset to dump gcc trees to text and restore them from text, to make debugging as well as sneaking at other compilers possible, and to avoid recompiling GCC to do the job.

Another way would be to make our back-end output C which could then be compiled by GCC or even other compilers. This could be even better.

-eye.

In article <bor5gt$1uv0$1@digitaldaemon.com>, Ilya Minkov says...
>
>I'm looking for a suitable toolset to dump gcc trees to text and restore them from text, to make debugging as well as sneaking at other compilers possible, and to avoid recompiling GCC to do the job.

Have you looked at Yaml? (www.yaml.org)

It's supposed to be as good as XML, but a lot easier to parse, and it definitely is totally readable, vs XML which I find totally unreadable.

While Yaml is not a toolset, it is meant to be easy enough to both emit and parse that you can code your own toolset in a matter of a couple of days.

Actually, I've been toying with the idea of using Yaml to serialise D objects.

Georg Wrede wrote:
> Have you looked at Yaml? (www.yaml.org)

I know Yaml. The problem is not serialising and reading. One problem is understanding the GCC side.

And THE problem is that i probably have *no* time at all until summer...

I'm sorry that i can't hold promises.

-eye