Florian Heigl wrote:
> (now if I just managed sending this to the list in the first place.)
>
> ---------- Forwarded message ----------
>
> Hi list,
>
> just a few notes for the curious :)
> I also have mirrored Tobias Nygren's old pkgsrc and notes, he tarred
> them up for me last year; but I'll try to rebuild stuff using the new
> diff.
> http://wartungsfenster.dyndns.org/outbox/index.php?path=hpux_stuff/
>
> Be warned: Using gcc on hp-ux might break pkgsrc, and using HP's ansi
> c compiler will
> break some GNUish software [ on 11.11 this includes things like bash ] :)
> Also it's very healthy to double-check your /usr/pkg/bin/libtool.
>
> I have compiled a list of the fancy bugs I hit over the last weeks at
> http://justblog.blogdns.org/netzwerk/stories/2955/
> it's in german, but I'll translate it back to english later tonight.
>
> If you can actually produce working code (I can't), i think the two
> best things would be making pkgsrc to produce a working bootstrap
> and/or port the bulk-sandbox script for use on hp-ux. The latter would
> probably even better, as it would allow to rebuild things using the
> pkgsrc gcc.
>
> Last, there is definitely demand for pkgsrc on hp-ux, as there is only
> one large source of opensource software for HP-UX and it has it's
> 'flaws'. I know at least 10 people that would like to run pkgsrc
> instead, not to mention 150 hungry boxes @work.
> Our policy doesn't really allow using precompiled software, but
> compiling in-house is ok, so pkgsrc would be *really* helpful.
>
> regards and thanks to all involved,
> florian
>
> -- 
> 'Sie brauchen sich um Ihre Zukunft keine Gedanken zu machen'

Hello,

I hope you'll have better luck with the new "release".
the latest shells/bash does compile fine on 11.11 with HP's compiler.
You bug reports are valuable. I am mostly interested in
infrastructural bugs such as mk/* problems and breakage in perl,
libtool, openssl and such. The more exotic packages will come later.

The newer HP-UXen use ELF binaries instead of SOM. The
support for this should be easy to implement, but I really need a
shell account on an Itanium box to the the detection logic right.

I plan to run triweekly bulk builds on the hardware I have in my
possession, once the mk/* changes settle down.