Re: [AMBER-Developers] Should we add more troubleshooting to the Makefile?

From: David Case <>
Date: Mon, 6 Feb 2017 22:26:52 -0500

On Mon, Feb 06, 2017, Dave Cerutti wrote:
> Perhaps we should discuss ways to add features to the installation script
> that will detect common errors or vulnerabilities we can foresee, then
> prompt the user to take specific actions to rememdy them. Some of this is
> already in place, such as suggestions to source and
> when encountering certain problems. In the case of my python problem, it
> would have been helpful for the installation to show me my actual
> $PYTHONPATH. Should we make an effort to broaden the scope of these hints?

General agreement here, although puts $AMBERHOME/lib... first
in PYTHONPATH, so you presumably have some script that modifies PTYONPATH
after this, which means it will be tough for our scripts to catch that.

But overall, I'd like to push to make configure2 simpler, rather than
more complex. I argued for this a few weeks ago (maybe in offline emails
with Dan, Jason and Hai), without much success. But, just to get people
thinking in advance of the upcoming meeting:

1. There are 4 mic* options that require use of the 2012 or later Intel
   compilers, yet are still listed as "experimental". Can they be simplified?
   Same for SSE_TYPES: is anyone still using this?

2. Given that Windows10 has a new and functional Linux subsystem, do
   we need to continue to support the -cygwin, -wine and -windows options?

3. Does anyone use -static?

4. Does anyone use -nosse?

6. Do we need -nosanderapi?

7. Do we need to support all three options for --python-install? (This was
    the subject of our previous email thread.)

8. Is anyone using the -netcdfstatic option?

9. Do we have to support -dragonegg?


AMBER-Developers mailing list
Received on Mon Feb 06 2017 - 19:30:03 PST
Custom Search