www.digitalmars.com         C & C++   DMDScript  

D.gnu - Builded gdc 0.24 - 1 error - some questions

reply "Carlos" <carlos-smith sympatico.ca> writes:
Well, i am glad. I just builded gdc 0.24.
Just follwed the instructions, and every thing was ok.

I got one error during the build.

The "fix headers" or something like this, did not ran Ok.

The reason being that there was no /local/include directory.
So i just created one, et relaunched make.

==> Eveything looks ok, but is it ???

I have Msys and Mingw installed into different directories.
    E:\Msys
    E:\MingW

I did not installed gdc with Mingw.

If you dont specify any special options, the installation is
done in E:/Msys/local/...

I would also like to know:

==> Do we have to build the C and the C++ compiler. Can we 
just build the D compiler ?

==> What do i take in the trunk and where do i put it in order
to build SVN V2 ? (Ok, i did not look very hard!)


==> Is there some tech documentation on gdc ?
I would like to know where is the line between DMD code
and gcc code.
How do you interface gcc and dmd ?
Do you execute all the semantic phases of dmd ?

Please some hints woul help a lot.

I already made 2 little modification to the compiler to see how it goes.
Now i would like to do some more.

Thanks for any help.
Aug 21 2008
parent reply Mosfet <mosfet anonymous.org> writes:
Carlos wrote:
 Well, i am glad. I just builded gdc 0.24.
 Just follwed the instructions, and every thing was ok.
 
 I got one error during the build.
 
 The "fix headers" or something like this, did not ran Ok.
 
Could you give more information, GCC version, host system, configure parameters ...
Sep 03 2008
next sibling parent "Carlos" <carlos-smith sympatico.ca> writes:
"Mosfet" <mosfet anonymous.org> a écrit dans le message de news: 
g9ltjl$mhc$1 digitalmars.com...
: Carlos wrote:
: > Well, i am glad. I just builded gdc 0.24.
: > Just follwed the instructions, and every thing was ok.
: >
: > I got one error during the build.
: >
: > The "fix headers" or something like this, did not ran Ok.
: >
: Could you give more information, GCC version, host system, configure
: parameters ...
:
:
:
:
: 
Sep 03 2008
prev sibling parent "Carlos" <carlos-smith sympatico.ca> writes:
Oops, sorry...

: Could you give more information, GCC version, host system, configure
: parameters ...

Hi Mosfet!,

Thanks for responding.

I had to stop working on this project. But, this morning,
after reading your post, i decided to check it again.

I discovered why the script emit this error message:
  ==> constructing ../fixinc.sh for i686-pc-mingw32 to run on 
i686-pc-mingw32
  ==> The directory that should contain system headers does not exist:
  ==>   /usr/include

Somewhere during the build of gcc-3.4.5-core, it must be determined
if the headers of previous version of the compiler must be fixed.
For some machines, the headers must be fixed, for some others, they
are ok and dont need to be fixed.

After the check, there is a script: fixinc.sh which is created
(and contains a simple exit for MingW), because with MingW32
compiler on WinXP, the headers do not need fixing.

The script that determine that run ok and correctly find that there
is nothing to fix.

But, somewhere in the Makefile, after fixinc.sh has been created,
there is a target:

    stmp-fixinc: fixinc.sh gsyslimits.h
        ...
        ...
who check for the presence of /usr/include before trying to fix
any headers, whether there is something to fix or not. ( i thing they
sould avoid that).

The problem is, that /usr is mounted on E:/MSys, and there is
no such directory as /usr/include. That's because MSys is just the bare
minimum needed to execute configure. It does not, and should not
contains an include directory.

So, to fix this, knowing that MingW32 do not need fixing ( i did not
understood that when i posted the firts time),
i simply deleted a dozen of lines or so in the Makefile.

Hope i am clear. (My english is not up to the task...)

Carlos
Sep 03 2008