Re: ocaml build fix

[Jonathan Perkin <jperkin%joyent.com@localhost>]

* On 2017-03-27 at 07:43 BST, Joerg Sonnenberger wrote:

> On Sun, Mar 26, 2017 at 11:02:02PM -0400, Greg Troxel wrote:
> > 
> > Joerg Sonnenberger <joerg%bec.de@localhost> writes:
> > 
> > > On Sun, Mar 26, 2017 at 11:48:57AM -0400, Greg Troxel wrote:
> > >> ocaml has been broken for NetBSD i386 since before the freeze started.
> > >> I am unclear on if there is trouble on other platforms.  NetBSD amd64
> > >> and OS X both seem ok.  The problem is that the "raw_spacetime" library
> > >> is not built, but it's in the PLIST.  As far as I know, it's only this
> > >> PLIST mismatch that is wrong.  The patch below merely omits the
> > >> raw_spacetime libs from PLIST on NetbSD i386.
> > >
> > > I don't think this change makes sense and it is much more likely to hide
> > > the real issue. As such,  I am against this patch.
> > 
> > Do you think it's better that ocaml and all ocaml programs fail to build
> > for the branch?   As far as I can tell this is some build detection nit
> > and the library isn't being built for some not particularly good reason.
> 
> Compared to a patch that silently ignores a wrong OS being picked up?
> Actually, yes, I do. I had too much fun dealing with ABI misselection
> and things like that to be happy with this patch...

Note that SmartOS/i386 fails in the same way:

  http://us-east.manta.joyent.com/pkgsrc/public/reports/upstream-trunk32/20170325.2239/ocaml-4.04.0/install.log

However even with that, both i386 and x86_64 would still fail due to
the std_exit and stdlib diffs:

  http://us-east.manta.joyent.com/pkgsrc/public/reports/upstream-trunk64/20170324.2240/ocaml-4.04.0/install.log

So, I'm not sure it's completely down to OS mis-selection, but I do
think we should come up with a decent way to handle these changes, as
they break on every ocaml update.

-- 
Jonathan Perkin  -  Joyent, Inc.  -  www.joyent.com