From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me
X-Spam-Level: 
X-Spam-Status: No, score=0.4 required=5.0 tests=BAYES_00,FORGED_MUA_MOZILLA
	autolearn=no autolearn_force=no version=3.4.4
X-Google-Thread: 103376,5f6125e9594f9968
X-Google-NewGroupId: yes
X-Google-Attributes: gida07f3367d7,domainid0,public,usenet
X-Google-Language: ENGLISH,ASCII-7-bit
Received: by 10.68.231.138 with SMTP id tg10mr19429415pbc.7.1332771242485;
        Mon, 26 Mar 2012 07:14:02 -0700 (PDT)
Path: 
 z9ni3893pbe.0!nntp.google.com!news1.google.com!news3.google.com!fu-berlin.de!uni-berlin.de!individual.net!not-for-mail
From: Niklas Holsti <niklas.holsti@tidorum.invalid>
Newsgroups: comp.lang.ada
Subject: Re: Ada to C translator for small microcontrollers
Date: Mon, 26 Mar 2012 17:14:00 +0300
Organization: Tidorum Ltd
Message-ID: <9tbbt9Fhe9U1@mid.individual.net>
References: <2424259e-cf1c-408e-a57e-f2c948087186@i2g2000vbv.googlegroups.com>
Mime-Version: 1.0
X-Trace: individual.net RIO7oERZoR/D44jmO3MX+g4diEn8LQlO1CNmwPi84lkmFHfR1Q
Cancel-Lock: sha1:Y1OFzpAbP6HeST0HlSxPuOpC7wE=
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.6;
 rv:10.0.2) Gecko/20120216 Thunderbird/10.0.2
In-Reply-To: 
 <2424259e-cf1c-408e-a57e-f2c948087186@i2g2000vbv.googlegroups.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Date: 2012-03-26T17:14:00+03:00
List-Id: <comp.lang.ada>

On 12-03-26 15:48 , Tomi Saarnio wrote:
> Hi all,
>
> As there exists at least one free decent C cross-compiler for smaller
> 8-bit and 16-bit architectures (Pic, Avr, 8051 etc.), but none Ada
> compiler that I know of,

A partial port of GNAT for the AVR exists: 
sourceforge.net/apps/mediawiki/avr-ada/.

> I was wondering whether there is a demand for
> an Ada to C translator, that would implement some sort of Ada subset
> and output corresponding ANSI C source code.

SofCheck (www.sofcheck.com) provides such an Ada (95) compiler, a 
version of their AdaMagic front-end.

> I am not a compiler or
> Ada expert, so I cannot really estimate the effort how hard this would
> be.

Quite large, I think, unless you take a very small subset. But one would 
really like the subset to provide most of the compile-time advantages of 
Ada: packages, user-defined types, type attributes, generics.

I think the realistic approach for an open-source implementation would 
be to take the existing GNAT front-end and try to generate C code from 
the intermediate representation, either the gcc IR or the llvm IR.

> As far as I can see the problem, the first thing is to identify a
> suitable subset(s) to be implemented, and then to come up with the
> corresponding C source idioms.

In the GNAT-based approach that I suggest above, the bottleneck is in 
the IR-to-C translator. I don't know much about the gcc or llvm IRs, but 
I would expect the subset to be defined by the limitations of whatever 
IR-to-C translator is created, and it may not be easy to define the 
corresponding subset on the Ada level. In any case, it seems simpler to 
start building an IR-to-C translator and accept whatever limitations it 
turns out to have.

There are some "Ada to C/C++" translators that work on the pure 
source-to-source "idiom" basis, but I believe they are intended to help 
porting Ada projects from Ada to C or C++, and probably require manual 
assistance to finish the translation. The SofCheck tool is a fully 
automatic, real compiler that generates complete and finished C source.

-- 
Niklas Holsti
Tidorum Ltd
niklas holsti tidorum fi
       .      @       .