Re: [AMBER-Developers] Amber12 release candidate 2

From: Lin Fu <linfu3200.gmail.com>
Date: Thu, 22 Mar 2012 15:08:57 -0400

*Not sure why the parameter folder ($AMBERHOME/dat/mtkpp) for mtkpp is gone
in current candidate package*. If this folder is added for mtkpp, The test
cases for mtkpp on the cygwin will pass except two. After checking with
these two, actually one difference is due to round error, the other is
different order of atom number for given ring in the molecule detected by
the ringTest.exe, which probably won't affect the result. but not sure
whether Labute algorithm for ring dection will give different order of atom
number in different platforms, is that normal or expected.

Best Regards,
Fu Lin
---------- Forwarded message ----------
From: Lin Fu <linfu3200.gmail.com>
Date: 2012/3/22
Subject: Re: [AMBER-Developers] Amber12 release candidate 2
To: AMBER Developers Mailing List <amber-developers.ambermd.org>


Hi Martin,

The mtkpp can be compiled successfully on the cygwin, as long as "#include
<time.h>" has been added in the some cpp files under the dir of
mtkpp/tests. But I don't have time to test all the binary generated from
mtkpp since the cygwin on my laptop is quite slow.

Best Regards,
Fu Lin

2012/3/22 Martin Peters <martin.b.peters.me.com>

> Hi Lin,
>
> Did you test that fix? Did mtkpp compile? If so, could someone with the
> necessary write privileges make the change?
>
> Cheers,
> Martin
>
> On 22 Mar 2012, at 01:19, Lin Fu wrote:
>
> > The problem is variable $host in the configure file. In the cygwin, the
> > host is "i686-pc-cygwin", rather than "i686-pc-linux-gnu" or
> > "x86_64-unknown-linux-gnu" in the common Linux OS. Therefore, the "case"
> > selection will always go to the last option
> > with "platform=UNKNOW". If adding extra condition for cygwin, it will
> > configure properly.
> >
> > case "${host}" in
> > *-*-linux* | **-*- cygwin** ) platform=LINUX
> > tmp_FC='ifort ifc pgf90 xlf90_r f90'
> > tmp_CC='gcc'
> > tmp_CXX='gcc'
> > is_linux=yes
> >
> >
> >> vim configuration
> > ---------------------
> > ---------------------
> > ---------------------
> > 16623 case "${host}" in
> > 16624 *-*-linux*) platform=LINUX
> > 16625 tmp_FC='ifort ifc pgf90 xlf90_r f90'
> > 16626 tmp_CC='gcc'
> > 16627 tmp_CXX='gcc'
> > 16628 is_linux=yes
> > 16629
> > 16630 $as_echo "#define HOST_LINUX 1" >>confdefs.h
> > 16631
> > 16632 ;;
> > 16633 *-apple-darwin*) platform=MACINTOSH
> > 16634 tmp_FC='f90 xlf90_r'
> > 16635 tmp_CC='gcc'
> > 16636 tmp_CXX='gcc'
> > 16637 is_darwin=yes
> > 16638 if test "${bMKL1}" = "no" -o "${bMKL2}" = "no" ; then
> > 16639 APPLE_LDFLAGS="-framework Accelerate"
> > 16640 else
> > 16641 APPLE_LDFLAGS=""
> > 16642 fi
> > 16643 APPLE_LDFLAGS="$APPLE_LDFLAGS -static"
> > 16644 APPLE_LIBS="-lz"
> > 16645 LDFLAGS="$APPLE_LDFLAGS $LDFLAGS"
> > 16646 LIBS="$APPLE_LIBS $LIBS"
> > 16647
> > 16648 $as_echo "#define HOST_DARWIN 1" >>confdefs.h
> > 16649
> > 16650 ;;
> > 16651 *) platform=UNKNOWN
> > 16652 is_unknown=yes
> > 16653 echo 'ERROR: this platform is not supported at this time!'
> > 16654 exit ;;
> > ---------------------
> > ---------------------
> > ---------------------
> >
> > 2012/3/21 Ben Roberts <ben.roberts.geek.nz>
> >
> >> I've also forwarded the relevant parts of this thread to Martin Peters,
> >> who can perhaps put something in the configure scripts to fix it more
> >> properly, e.g., by returning a non-zero exit code.
> >>
> >> On 22/03/2012, at 10:02 AM, Daniel Roe wrote:
> >>
> >>> Same issue in configure.in (line 386). I think the bottom line is,
> >>> unless someone knows that mtkpp should/does work under cygwin, it
> >>> should be disabled when '-cygwin' is specified.
> >>>
> >>> -Dan
> >>>
> >>> On Wed, Mar 21, 2012 at 2:46 PM, Jason Swails <jason.swails.gmail.com>
> >> wrote:
> >>>> But configure is automatically generated by autoconf, yes? If so, is
> >> this
> >>>> an issue with autoconf or the configure.in file?
> >>>>
> >>>> On Wed, Mar 21, 2012 at 4:39 PM, Daniel Roe <daniel.r.roe.gmail.com>
> >> wrote:
> >>>>
> >>>>> More details: MTKPP configure actually does choke under cygwin, but
> >>>>> exits with no status (hence 0), so Amber configure doesnt think
> >>>>> anything is wrong.
> >>>>>
> >>>>> The issue is on line 16654 of mtkpp/configure:
> >>>>>
> >>>>> *) platform=UNKNOWN
> >>>>> is_unknown=yes
> >>>>> echo 'ERROR: this platform is not supported at this time!'
> >>>>> exit ;;
> >>>>>
> >>>>> That should be changed to 'exit 1'.
> >>
> >>
> >> _______________________________________________
> >> AMBER-Developers mailing list
> >> AMBER-Developers.ambermd.org
> >> http://lists.ambermd.org/mailman/listinfo/amber-developers
> >>
> > _______________________________________________
> > AMBER-Developers mailing list
> > AMBER-Developers.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber-developers
>
>
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber-developers
>
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Thu Mar 22 2012 - 12:30:02 PDT
Custom Search