Mail Archives: djgpp-workers/2023/09/05/13:17:52
On 9/5/23 03:35, Rugxulo (rugxulo AT gmail DOT com) [via djgpp-workers AT delorie DOT com] wrote:
> Hi,
>
> On Mon, Sep 4, 2023 at 11:29 AM Andris Pavēnis (andris DOT pavenis AT iki DOT fi)
> [via djgpp-workers AT delorie DOT com] <djgpp-workers AT delorie DOT com> wrote:
>> I finally uploaded a month old build of gcc-13.2.0 to http://www.delorie.com/pub/djgpp/beta/v2gnu/
> So this really is "beta"?? (I guess you still want further testing.)
>
> Is it still using stock 2.05 or newer from CVS?
>
>> I have not done real tests with:
>>
>> - No Ada compiler (do not expect one from me)
> Understandable, I don't expect it, but you did here include "ada132d.zip" docs.
My omission. Script for building docs is separate from building cross- or native-compiler. I simply
saw that script succeeds and uploaded without too much additional thinking.
> BTW, why no attempt at Modula-2? (No offense, I know time and energy
> are probably low.)
That did not even to come into mind. Anyway is easy to try to build cross-compiler under ArchLinux.
Tried (build took about 5 minutes on Ryzen 9 3950X CPU). It does not however find SYSTEM.mod, which
is installed when tried to compile an example which I found somewhere on web (I do not know
language myself).
If You know language (Modula 2) and have access to ArchLinux: You could take files from
https://aur.archlinux.org/packages/djgpp-gcc (https://aur.archlinux.org/djgpp-gcc.git) and do
following additional changes:
[andris AT ap djgpp-gcc]$ git diff
diff --git a/PKGBUILD b/PKGBUILD
index ade056d..b518f00 100644
--- a/PKGBUILD
+++ b/PKGBUILD
@@ -7,7 +7,7 @@pkgver=13.2.0
_target="i686-pc-msdosdjgpp"
_djver=2.05
_build_ada=no
-pkgrel=2
+pkgrel=3
pkgdesc="GCC for the djgpp cross-compiler"
arch=('i686' 'x86_64')
url="http://gcc.gnu.org"
@@ -40,10 +40,10 @@build() {
  rm -rf gcc-build-native gcc-install-native gcc-build-$_target
  if [ "$_build_ada" == "yes" ] ; then
     _bootstrap_languages=c,c++,ada
- Â Â Â Â _build_languages=c,c++,fortran,objc,obj-c++,ada
+ Â Â Â Â _build_languages=c,c++,fortran,objc,obj-c++,ada,m2
  else
     _bootstrap_languages=c,c++
- Â Â Â Â _build_languages=c,c++,fortran,objc,obj-c++
+ Â Â Â Â _build_languages=c,c++,fortran,objc,obj-c++,m2
  fi
  if [ "$(gcc -dumpversion | sed -e 's:\..*::')" != "$(echo $pkgver | sed -e 's:\..*::')" ] ; then
     echo "Different GCC major version: building native compiler at first"
Perhaps the same could also be done for RPM based systems.
>
>> - the GCC bootstrap (https://gcc.gnu.org/install/build.html) succeeded (so it works that far)
>>
>> - rebuilt GMP-6.2.0 (no problems encountered)
>>
>> - building not very large C++ program (the program that generates installation ZIP archives is now
>> C++ and uses C++20) - it created binary archives after some small changes
> Should I try to rebuild p7zip (C++) with it?
You could try and tell about results.
>> - no testing on non-LFN systems at all (but earlier mentioned C++ program incorporates the
>> functionality of doschk in slightly more advanced form and refuses to create ZIP packages in case
>> of a non-LFN conflict)
> Dumb question, but what OS is expected to run it? WinXP? Vista? 10? FreeDOS?
It is supposed to work in listed ones. I have not tried FreeDOS 1.3 myself, but I had bad
experience with LFN support in FreeDOS 1.2. So You are welcome to try.
Should perhaps work also in Windows 7 32 bit, maybe Windows 8 32 bit and even earlier versions of
WIndows 10 32 bit before Microsoft broke DPMI support.
>
>> - no testing at all for such new C++20 feature as C++ modules - they may not work at all. I do not
>> know. And it is highly unlikely that I'll try to fix it if it does not work
> Modules were a huge change, and I don't think GCC fully supports them
> yet anyways.
>
I had to exclude several ways of module support (only one left) as they were unusable for DJGPP
Andris
- Raw text -