Date: Sun, 24 Oct 2010 09:27:49 -0400 (EDT)
From: Jun-Hui Zhao 
To: Dan Marrone 
Cc: Peter Williams ,
     Peter Teuben ,
     Colby Gutierrez-Kraybill ,
     smamiriad@cfa.harvard.edu
Subject: Re: miriad beta trouble

Hi Dan and Peter,
      Thanks. Please let me know any detailed changes that you made
in compiling Miriad on Mac. Pat has submitted a perchase request.
I think that we will get a Mac here and will incorporate the
corrections into the SMA Miriad code for the final release (Merging
back to Miriad CVS).
      Jun-Hui
On Sun, 24 Oct 2010, Dan Marrone wrote:

> Hi all,
>
> Peter, thanks for taking the time to look at this in enough detail to get it
to compile. I also got it to compile, thanks to your patch, through the
install.miriad route. unfortunately, the crucial smalod program crashes
immediately. The first problem is that all the 'version' strings are filled via
sprintf with strings longer than the defined length of the variable. Once
fixed, smalod will run but not do anything. I haven't yet had time to look into
this further. It's not clear to me why the program (at least in the 13Oct
tarball) should work in linux and fail this way in osx, but now that I'm on my
way back from Chile perhaps I can try compiling this updated version (18Oct) on
my linux desktop to see if it also fails there.
>
>
> I think this effort has gotten the mac route closer to functionality, so that
is helpful. If the reintegration of this code into the main distribution is
proceeding, I expect that will be the ultimate way to ensure that the SMA data
is properly supported. However, if it can be made to work on a shorter
timescale through a slightly more complicated path (presently patching and
merging by hand), that would be even better. I'll keep playing with it when I
have an opportunity.
>
> Dan
>
> On Oct 18, 2010, at 8:24 PM, Peter Williams wrote:
>
> > On Mon, 2010-10-18 at 17:30 -0400, Peter Teuben wrote:
> >> Here's another idea:     i'm a tiny bit worried that the merge may break
the
> >> current build, but you're a better judge than me.
> >> Instead of overlaying your tar file on a 4.1.7 tree, perhaps you should
> >> cvs checkout the current (*i.e. 4.1.7) tree, and branch it, and overlay
> >> the tar in that one, do you merge work  (and you're free to now commit
> >> those as you please, so we can all see those). Then we confirm the old
> >> and new style build work. If that's done, we merge the tree back.
> >
> > To aid in understanding the SMA changes, I've ported them onto a local
> > copy of the current CVS HEAD, resolved a few (insignificant) conflicts
> > in the merge, and tweaked the autotools build system to build them
> > correctly. The resulting diffs are here:
> >
> >  http://astro.berkeley.edu/~pkwill/sma-integration.diff.gz (86K)
> >  http://astro.berkeley.edu/~pkwill/sma-full.diff.gz (1.3M)
> >
> > The former diff omits some large data files added in $MIRCAT, changes
> > within src/prog/sma/, and whitespace-only changes. The latter diff
> > contains everything.
> >
> > I tried building the resulting source tree on a Mac OS X laptop and
> > found one problem that broke the build: a few symbols were being
> > duplicated in some of the SMA subroutines. Attached is a patch that
> > fixes this problem. (That is, things build; I haven't tested if they
> > work.)
> >
> > Peter
> >
> > --
> > Peter Williams / pwilliams@astro.berkeley.edu
> > Department of Astronomy, UC Berkeley
> > 
>