On 2012-02-13 22:41, Leandro Lucarella wrote:
> Johannes Pfau, el 13 de febrero a las 20:57 me escribiste:
>>> The problem with this is, if you are building gdc, you have to
>>> clone/download THREE repositories/tarballs and move/extract them to
>>> the correct places, which are in varying locations in the gcc source
>>> code.
>>
>> Git submodules might be useful here, that's what the ldc
>> developers do. See
>> https://github.com/ldc-developers/ldc/tree/master/runtime
>> but I have to admit I don't know anything about git submodules.
>
> Git submodules can definitely take care of this. Of course is a little
> more tedious to use than having just a plain monolithic repo, but it's
> not that bad.
>

I'm not sure they will help if you have to do this:

"and move/extract them to the correct places, which are in varying locations in the gcc source code"

It might help somewhat but not completely.

-- 
/Jacob Carlborg

Am Mon, 13 Feb 2012 19:30:07 +0000
schrieb Iain Buclaw <ibuclaw@ubuntu.com>:

> On 13 February 2012 18:46, Johannes Pfau <nospam@example.com> wrote:
> > Am Mon, 23 Jan 2012 20:13:26 +0000 (UTC)
> > schrieb Iain Buclaw <ibuclaw@ubuntu.com>:
> >
> >> Started an initial list for our roadmap for the next month.
> >>
> >> https://bitbucket.org/goshawk/gdc/wiki/Roadmap
> >>
> >>
> >> Daniel, if there's anything you could help on, please don't hesitate to contact me. :)
> >>
> >>
> >> Regards
> >> Iain
> >
> > "Re-structure project directories to match that of GCC"
> >
> > can you give some more information about that? Will druntime, phobos1 and phobos2 all get own subdirectories in the gcc folder? I'll try to make druntime buildable out-of-tree, so it'd be good to know.
> >
> 
> Yeah, D1 and D2 will have to be split from each other, as the library folders share the same name, thus conflict.
> 
> The new directory structure will be:
> 
> /
> ... /gcc
> ... ... /d
> ... ... ... /dfrontend
> ... /libphobos
> ... ... /core
> ... ... /etc
> ... ... /gc
> ... ... /gcc
> ... ... /rt
> ... ... /std
> 

Is there a special reason why you want to keep druntime & phobos in one
directory? If you just don't have the time to split druntime & phobos,
I could do it. I've already partially done this when
adding support for out-of-tree builds to druntime (That's working now,
but I need a little more time to clean up the patches and test multilib)

On 15 February 2012 18:21, Johannes Pfau <nospam@example.com> wrote:
> Am Mon, 13 Feb 2012 19:30:07 +0000
> schrieb Iain Buclaw <ibuclaw@ubuntu.com>:
>
>> On 13 February 2012 18:46, Johannes Pfau <nospam@example.com> wrote:
>> > Am Mon, 23 Jan 2012 20:13:26 +0000 (UTC)
>> > schrieb Iain Buclaw <ibuclaw@ubuntu.com>:
>> >
>> >> Started an initial list for our roadmap for the next month.
>> >>
>> >> https://bitbucket.org/goshawk/gdc/wiki/Roadmap
>> >>
>> >>
>> >> Daniel, if there's anything you could help on, please don't hesitate to contact me. :)
>> >>
>> >>
>> >> Regards
>> >> Iain
>> >
>> > "Re-structure project directories to match that of GCC"
>> >
>> > can you give some more information about that? Will druntime, phobos1 and phobos2 all get own subdirectories in the gcc folder? I'll try to make druntime buildable out-of-tree, so it'd be good to know.
>> >
>>
>> Yeah, D1 and D2 will have to be split from each other, as the library folders share the same name, thus conflict.
>>
>> The new directory structure will be:
>>
>> /
>> ... /gcc
>> ... ... /d
>> ... ... ... /dfrontend
>> ... /libphobos
>> ... ... /core
>> ... ... /etc
>> ... ... /gc
>> ... ... /gcc
>> ... ... /rt
>> ... ... /std
>>
>
> Is there a special reason why you want to keep druntime & phobos in one
> directory? If you just don't have the time to split druntime & phobos,
> I could do it. I've already partially done this when
> adding support for out-of-tree builds to druntime (That's working now,
> but I need a little more time to clean up the patches and test multilib)
>

That is how it currently looks in the GCC source directories once you get past the ./setup-gcc.sh stage.  It would be great if druntime was a subdirectory, so if you have time to do this before I get round to it, patches to the configure scripts are always welcome. :-)


-- 
Iain Buclaw

*(p < e ? p++ : p) = (c & 0x0f) + '0';