Re: Issues with compiler failures

[Anders Magnusson <ragge%ludd.ltu.se@localhost>]

Den 2019-03-22 kl. 21:51, skrev Johnny Billquist:
> On 2019-03-22 18:04, Silas wrote:
> > On Sun, Mar 17, 2019 at 08:07:55AM +0100, Martin Husemann wrote:
> > > An easier option would be to make pcc usable and accept it leads to 
> > > some
> > > slightly reduced functionality (i.e. no groff, atf and libexecinfo in
> > > the base build due to missing c++ support).
> > This sounds interesting.  I just download pcc-20170101 (the latest?) 
> > and there is an arch/vax directory with stuff within. The $Id$ 
> > keyword of arch/vax/code.c  even blames ragge for the latest commit :-)
> > Isn't this usable?
> As far as I can remember, there are still some issues meaning it might 
> not compile everything under NetBSD correctly. Not to mention the lack 
> of C++.
Don't try C++.
The last bugs I was dealing with were related to long long, the other 
code in pcc has been there since -79, or so, so it is most likely useful 
even today.
I'm currently dealing with using softfloat routines in pcc, so that 
cross-compiling works.  It's incredible much faster to try things on a 
new machine than on a 4000/90 :-)
-- R