Re: Clang/LLVM migration roadmap

[Riccardo Mottola]

Hi Andreas,

I have around 40 different computers where I run GNUstep on. So
something to play&test :)


Andreas Fink wrote:
> What are the real issues of the runtime on other platforms?

They must be checked 1:1, maybe I could setup a wiki page with "tested"
support.

E.g. NetBSD/i386 is not working due to completely broken exceptions.
FreeBSD/amd64 has broken exception, they do not catch.

On other lesser known platforms I desisted, for now, but can set up.
Perhaps a wiki page with a matrix of the status.

The nuiseance is that I do not have found a reliable method to have two
installs at the same time, if of one machine setup I have only one.
Maybe I could have to different roots like /usr/GNUstep-GCC and
/usr/GNUtep-CLang and then source the scripts. But also everything needs
to be compiled twice... two installs, two source codes! Will need to test.

> I might have a personal interest to do some work on some platforms.

We have two level of issues. Availability of clang and working of the
runtime itself.


I did a quick check for NetBSD and the availability of CLang as a package.

Some notable omissions which would affect me:
* PowerPC
* MIPS-EL
* 68K

Sparc & sparc64 seem to have clang! As far as I know it has issues, but
I will check! However then we need to have the libobjc2 runtime support
it, but that is "easier" since immediately in our hands.

I'd like to get the hands dirty on SPARC and PPC again in libobjc2... we
could start from linux if not on NetBSD.

Riccardo