GDC 0.17 is unfortunately out of date with respect to DMD.  I've been hacking at it and got a working GDC based on DMD 0.154, simply by incrementally going version-by-version, repatching it, and making sure it still compiles.  It seems to be working.  Most importantly, it compiles much of the stuff that GDC 0.17 doesn't due to language changes.

I don't want to fork or anything ... if the changes are relatively working, it'd be nice to get them into GDC proper.

Also, I fixed a strange bug that was annoying me.  It seems that GDC 0.17 doesn't compile std.c.linux.linux in, even on a GNU/Linux system. The function is replicated by std.c.unix.unix, but unfortunately if a program does:

version (linux) { std.c.linux.linux... }

it would fail.  I dummied down std.c.linux.linux to not conflict with the other built-in things (and to import them of course), and it seems to work - at least I can finally build 'build' ^^


Another issue - libgphobos.a.  I don't quite understand why GDC 0.17 only makes a .a.  It makes binaries quite big.  I modified phobos' auto* stuff to use libtool and make libgphobos.so.0.  That too seems to work.

If anybody wants a copy, I can get it to you.  David: What's your status?  Are you still maintaining GDC?  Do you have an accessible public repository?  It'd be nice if I could make changes without feeling like I was working against you :(

 - Gregor Richards

Gregor Richards wrote:

> GDC 0.17 is unfortunately out of date with respect to DMD.  I've been
> hacking at it and got a working GDC based on DMD 0.154, simply by
> incrementally going version-by-version, repatching it, and making
> sure it still compiles.  It seems to be working.  Most importantly,
> it compiles much of the stuff that GDC 0.17 doesn't due to language
> changes.

I think Brad did something similar, so you might want to check with him.

And I've put all the DMD diffs up at http://www.algonet.se/~afb/d/diffs/

> Another issue - libgphobos.a.  I don't quite understand why GDC 0.17 only makes a .a.  It makes binaries quite big.  I modified phobos' auto* stuff to use libtool and make libgphobos.so.0.  That too seems to work.

DMD does it this way too (libphobos.a), for the moment it's a good way to make sure the programs run since Phobos is in a constant state of flux... Eventually I'm sure something like "libstdc++" will emerge ?
(hopefully something less of a DLL Hell then what that one is, but)

> If anybody wants a copy, I can get it to you.

I'm interested, email me where it can be found.

--anders

Gregor Richards wrote:

> GDC 0.17 is unfortunately out of date with respect to DMD.  I've been hacking at it and got a working GDC based on DMD 0.154, simply by incrementally going version-by-version, repatching it, and making sure it still compiles.  It seems to be working.  Most importantly, it compiles much of the stuff that GDC 0.17 doesn't due to language changes.
> 
> I don't want to fork or anything ... if the changes are relatively working, it'd be nice to get them into GDC proper.
> 
> Also, I fixed a strange bug that was annoying me.  It seems that GDC 0.17 doesn't compile std.c.linux.linux in, even on a GNU/Linux system. The function is replicated by std.c.unix.unix, but unfortunately if a program does:
> 
> version (linux) { std.c.linux.linux... }
> 
> it would fail.  I dummied down std.c.linux.linux to not conflict with the other built-in things (and to import them of course), and it seems to work - at least I can finally build 'build' ^^
> 
> 
> Another issue - libgphobos.a.  I don't quite understand why GDC 0.17 only makes a .a.  It makes binaries quite big.  I modified phobos' auto* stuff to use libtool and make libgphobos.so.0.  That too seems to work.
> 
> If anybody wants a copy, I can get it to you.  David: What's your status?  Are you still maintaining GDC?  Do you have an accessible public repository?  It'd be nice if I could make changes without feeling like I was working against you :(
> 
>   - Gregor Richards
I'm also interested ... I'm currently trying to get it
working with gcc 4.1.0 ...
Mathieu

I'm trying to upload my patch to the D Journal, the owner gave me access but hasn't quite hacked it up so I can upload it properly.  If anybody wants a direct mail of it, email me at Richards@codu.org

Also, I'm working on 0.156, should have that soon.

 - Gregor Richards

Gregor Richards wrote:
> I'm trying to upload my patch to the D Journal, the owner gave me access but hasn't quite hacked it up so I can upload it properly.  If anybody wants a direct mail of it, email me at Richards@codu.org
> 
> Also, I'm working on 0.156, should have that soon.
> 
>  - Gregor Richards

void[] on IRC gave me some space to upload it:

http://www.dprogramming.com/~gregorr/gdc.php

 - Gregor Richards

Gregor Richards wrote:

> void[] on IRC gave me some space to upload it:
> http://www.dprogramming.com/~gregorr/gdc.php

Thanks, got it... Will look at it later.

--anders

Gregor Richards wrote:

> GDC 0.17 is unfortunately out of date with respect to DMD.  I've been hacking at it and got a working GDC based on DMD 0.154, simply by incrementally going version-by-version, repatching it, and making sure it still compiles.  It seems to be working.  Most importantly, it compiles much of the stuff that GDC 0.17 doesn't due to language changes.

Okay, I got your updated version ("gdc-Ego") and unpacked it.

You didn't say if you had been merging with what Brad did for
"gdc-0.18-alpha1", so I will initially assume that you didn't.

So now we're looking at integrating:
gdc-0.17        (DMD 0.140)
gdc-0.18-alpha1 (DMD 0.148)
gdc-Ego         (DMD 0.156)

Plus whatever changes that David has done for the gdc 0.18
release already, such as for Linux PowerPC or 64-bit support.

Will run some diffs and then try to compile this for the Mac.

--anders

PS.
The only differences between DMD 0.155 and 0.156 are prints.
(some debugging printfs were not commented out, expression.c)

Anders F Björklund wrote:
> Gregor Richards wrote:
> 
>> GDC 0.17 is unfortunately out of date with respect to DMD.  I've been hacking at it and got a working GDC based on DMD 0.154, simply by incrementally going version-by-version, repatching it, and making sure it still compiles.  It seems to be working.  Most importantly, it compiles much of the stuff that GDC 0.17 doesn't due to language changes.
> 
> 
> Okay, I got your updated version ("gdc-Ego") and unpacked it.
> 
> You didn't say if you had been merging with what Brad did for
> "gdc-0.18-alpha1", so I will initially assume that you didn't.

That would be the wrong assumption, actually I did :)

> 
> So now we're looking at integrating:
> gdc-0.17        (DMD 0.140)
> gdc-0.18-alpha1 (DMD 0.148)
> gdc-Ego         (DMD 0.156)

gdc-Ego is the combination of 0.17, 0.18-alpha1 and DMD-0.156

> 
> Plus whatever changes that David has done for the gdc 0.18
> release already, such as for Linux PowerPC or 64-bit support.

Assuming he comes out of hibernation :(

> 
> Will run some diffs and then try to compile this for the Mac.

That'd be awesome.

> 
> --anders
> 
> PS.
> The only differences between DMD 0.155 and 0.156 are prints.
> (some debugging printfs were not commented out, expression.c)

Yeah, I know, but from 0.154->0.155 was huge.

 - Gregor Richards

Gregor Richards wrote:

>> You didn't say if you had been merging with what Brad did for
>> "gdc-0.18-alpha1", so I will initially assume that you didn't.
> 
> That would be the wrong assumption, actually I did :)

Excellent, I had reorganized it to match gdc-0.17, so I hadn't
run the diffs yet. All good then, we should be all synched :-)

> gdc-Ego is the combination of 0.17, 0.18-alpha1 and DMD-0.156

Great, would it be possible to reorganize it to match gdc-0.17,
and perhaps rename it to "0.18-alpha2" or something similar ?

>> Plus whatever changes that David has done for the gdc 0.18
>> release already, such as for Linux PowerPC or 64-bit support.
> 
> Assuming he comes out of hibernation :(

If he doesn't, we just have to ignore those two fringe targets...
I don't see anything wrong with us releasing a gdc 0.18 anyway ?

>> The only differences between DMD 0.155 and 0.156 are prints.
>> (some debugging printfs were not commented out, expression.c)
> 
> Yeah, I know, but from 0.154->0.155 was huge.

Yup, http://www.algonet.se/~afb/d/diffs/dmd-154-to-155-src.diff.gz
Just thought I'd mention the 156, since there was no "D Change Log"

--anders

Anders F Björklund wrote:
> Gregor Richards wrote:
> 
>> gdc-Ego is the combination of 0.17, 0.18-alpha1 and DMD-0.156
> 
> Great, would it be possible to reorganize it to match gdc-0.17,
> and perhaps rename it to "0.18-alpha2" or something similar ?

Probably a good idea.

Though I don't like that the version numbering scheme of GDC has no connection to the DMD frontend it uses ... guess that's just me :)

The reason I was organizing it like this is it keeps the middle IR-code stuff separate from the frontend stuff, but since that's not how it's compiled I guess that's not really a good choice X-P

>> Assuming he comes out of hibernation :(
> 
> 
> If he doesn't, we just have to ignore those two fringe targets...
> I don't see anything wrong with us releasing a gdc 0.18 anyway ?

Well, I've e-mailed him, hope to get a response one way or the other.

 - Gregor Richards