www.digitalmars.com         C & C++   DMDScript  

D.gnu - D for GCC?

reply Tapio Kelloniemi <spam09 thack.org> writes:
Hi

On www.opend.org it is said that port of D frontend to GCC's backends 
has started, but I can't find any other information. I would like to 
know, if anybody is doing it right now / any more? Also, if someone is 
doing it, what is the goal of this project? Just to port D frontend or 
perhaps to include it in the GCC distribution? Is there any information 
available about when this project might be in finished. I mean that is 
there any point to wait for GCC port or should I just search for some 
other language?

Thanks alot!

-- 
Tapio
May 05 2003
parent reply Stephan Wienczny <wienczny web.de> writes:
Tapio Kelloniemi wrote:
 Hi
 
 On www.opend.org it is said that port of D frontend to GCC's backends 
 has started, but I can't find any other information. I would like to 
 know, if anybody is doing it right now / any more? Also, if someone is 
 doing it, what is the goal of this project? Just to port D frontend or 
 perhaps to include it in the GCC distribution? Is there any information 
 available about when this project might be in finished. I mean that is 
 there any point to wait for GCC port or should I just search for some 
 other language?
 
 Thanks alot!
 

I don't think that someone is really working on it (correct me...). The goal of the project should be to have it in the GCC distribution. But do the GCC people want to have it? Walter is porting his backend to linux. CU Stephan
May 06 2003
parent reply Ilya Minkov <midiclub tiscali.de> writes:
Stephan Wienczny wrote:
 Tapio Kelloniemi wrote:
 
 Hi

 On www.opend.org it is said that port of D frontend to GCC's backends 
 has started, but I can't find any other information. I would like to 
 know, if anybody is doing it right now / any more? Also, if someone is 
 doing it, what is the goal of this project? Just to port D frontend or 
 perhaps to include it in the GCC distribution? Is there any 
 information available about when this project might be in finished. I 
 mean that is there any point to wait for GCC port or should I just 
 search for some other language?


Search for another language? NO WAY! :) ... unless you know any that *actually* suits you better.
 Thanks alot!

I don't think that someone is really working on it (correct me...).

Jan Knepper has been working on it - but doesn't seem to have gotten far due to lack of time.
 The goal of the project should be to have it in the GCC distribution. 
 But do the GCC people want to have it?

Who cares? Either they take it - IMO they probably will as long as it complies with specifications - all kinds of wierd compilers are already in the distro - and they have even been friendly to the developers... And if they don't - noone needs their permission to distribute a GCC-based compiler.
 Walter is porting his backend to linux.

Though linux is a major target, it's not everything. Walter's backend cannot cover non-x86 architectures, we need GCC for that. And i guess, BSD shall also stay uncovered for a while - unless a GCC port is done. -i.
May 06 2003
next sibling parent reply Stephan Wienczny <wienczny web.de> writes:
Ilya Minkov wrote:
 Tapio Kelloniemi wrote:

Though linux is a major target, it's not everything. Walter's backend cannot cover non-x86 architectures, we need GCC for that. And i guess, BSD shall also stay uncovered for a while - unless a GCC port is done. -i.

We should try to persuade a gcc developer to help us. gcc developer are the only ones who know how the language tree is created. Cu Stephan
May 06 2003
next sibling parent reply Ilya Minkov <midiclub 8ung.at> writes:
Stephan Wienczny wrote:
 We should try to persuade a gcc developer to help us. gcc developer are 
 the only ones who know how the language tree is created.

They are definately not the only ones - Jan Knepper does as well. :) Seriously, there has been some documentation. GnuPascal (IIRC) creators have hacked and documented tree creation. Just follow the links on the opend homepage. There have been a number of tutorials. I guess GnuPascal, GNAT, and other developers can also be of help. BTW, GnuPascal source itself could be interesting - we're dealing with a similar language. -i.
May 06 2003
parent Jan Knepper <jan smartsoft.us> writes:
Ilya Minkov wrote:

 Stephan Wienczny wrote:
 We should try to persuade a gcc developer to help us. gcc developer are
 the only ones who know how the language tree is created.


I am getting there... I converted "toy" from C to C++ in a couple of hours... I have some of the GLUE layer between the D-frontend and the GCC-backend done, but hesitate to put it out in it's current form as it is not quite working yet.
 Seriously, there has been some documentation. GnuPascal (IIRC) creators
 have hacked and documented tree creation. Just follow the links on the
 opend homepage. There have been a number of tutorials. I guess
 GnuPascal, GNAT, and other developers can also be of help.

Correct, there is some docco out there, but since I have been under serious time and financial constrains the last couple of months I have slowed down on the project... -- ManiaC++ Jan Knepper
May 06 2003
prev sibling parent reply Garen Parham <nospam garen.net> writes:
Stephan Wienczny wrote:

 We should try to persuade a gcc developer to help us. gcc developer are 
 the only ones who know how the language tree is created.
 

I've done some poking around, and in the newer 3.x releases there is a new way to tie in frontends that isn't yet documented (GIMPLE). It looks a lot cleaner than how you had to do it in 2.95, and if the IR can support C/C++/ObjC/Fortran/Ada that'd seem like a good sign that D should be doable. There are a couple of other projects ("open cobol" iirc :)) trying to make it in that way..so maybe someone interested could ask them for tips.
May 07 2003
parent reply Stephan Wienczny <wienczny web.de> writes:
Garen Parham wrote:
 Stephan Wienczny wrote:
 
 
We should try to persuade a gcc developer to help us. gcc developer are 
the only ones who know how the language tree is created.

I've done some poking around, and in the newer 3.x releases there is a new way to tie in frontends that isn't yet documented (GIMPLE). It looks a lot cleaner than how you had to do it in 2.95, and if the IR can support C/C++/ObjC/Fortran/Ada that'd seem like a good sign that D should be doable.

You could look at: http://gcc.gnu.org/ml/gcc/2002-08/msg01397.html
May 07 2003
parent reply Garen Parham <nospam garen.net> writes:
Stephan Wienczny wrote:


 You could look at:
 
 http://gcc.gnu.org/ml/gcc/2002-08/msg01397.html

Yup, thats it.
May 07 2003
next sibling parent Tapio Kelloniemi <spam09 thack.org> writes:
Garen Parham <nospam garen.net> wrote:
Stephan Wienczny wrote:


 You could look at:

 http://gcc.gnu.org/ml/gcc/2002-08/msg01397.html

Yup, thats it.

What about GCC 3.3? I saw something related to this(?) in the New Features list. -- Tapio
May 22 2003
prev sibling parent Mark T <Mark_member pathlink.com> writes:
In article <b9c9gb$m4s$1 digitaldaemon.com>, Garen Parham says...
Stephan Wienczny wrote:


 You could look at:
 
 http://gcc.gnu.org/ml/gcc/2002-08/msg01397.html

Yup, thats it.

it appears to be in the gcc 3.5 release http://www.google.com/search?hl=en&ie=UTF-8&oe=UTF-8&q=gimple+gcc&btnG=Google+Search
Jun 21 2003
prev sibling parent Jan Knepper <jan smartsoft.us> writes:
 Jan Knepper has been working on it - but doesn't seem to have gotten far
   due to lack of time.

Correct... Fortunately that might change once I have new income which hopefully will be soon. I had a followup interview today and I think it went well. When it comes through I will be back on D-GCC, especially for those platforms that the real D is not going to support.
 The goal of the project should be to have it in the GCC distribution.
 But do the GCC people want to have it?

complies with specifications - all kinds of wierd compilers are already in the distro - and they have even been friendly to the developers... And if they don't - noone needs their permission to distribute a GCC-based compiler.

Correct.
 Walter is porting his backend to linux.

cannot cover non-x86 architectures, we need GCC for that. And i guess, BSD shall also stay uncovered for a while - unless a GCC port is done.

Well, Walter might be happy enough to compile a BSD version as well. Also, BSD has such good Linux emulation that it might work anyways. -- ManiaC++ Jan Knepper
May 06 2003