ARRRGGHHHH!!! I knew I was not going mad!!! The CVS tree is all links to the
dcqtp directory so when I updated it in AMBER 11 it went and modified
everything in AMBER10 as well. Hence why the AMBER 10 tree no longer
represents the release version of the code.
If nobody objects to this I plan to remove all the dcqtp links and try to
move the AMBER10 tree back to being the AMBER10 release.
I am assuming that just deleting the amber10/src/dcqtp directory in the tree
and replacing it with ../../dcqtp will work fine? Does anybody know if this
is likely to cause problems?
Any better suggestions? I am assuming it will keep the history information
intact but maybe not. I will also remove all the links for AMBER11 and put
the dcqtp directory directly in the AMBER11 tree since past experience has
shown what a disaster links in the cvs tree can be.
Objections?
All the best
Ross
From: Ross Walker [mailto:ross.rosswalker.co.uk]
Sent: Friday, August 14, 2009 6:42 PM
To: 'matthew_seetin.urmc.rochester.edu'
Subject: RE: qmmm_div in amber10
Bummer... I wonder how I managed to do that. I have rolled it back to the
10.0 version please try updating the amber 10 tree now and see if it works.
Weird!
All the best
Ross
From: matthew_seetin.urmc.rochester.edu
[mailto:matthew_seetin.urmc.rochester.edu]
Sent: Friday, August 14, 2009 4:46 PM
To: Ross Walker
Subject: Re: qmmm_div in amber10
I think that something broken in the tree might be the case. The cvs log for
freq.F90 shows an update made in 10/2008, well after the actual release.
--Matt
Sent via BlackBerry from T-Mobile
_____
From: Ross Walker
Date: Fri, 14 Aug 2009 16:13:11 -0700
To: 'Seetin, Matthew'<Matthew_Seetin.URMC.Rochester.edu>
Subject: RE: qmmm_div in amber10
Hi Matthew,
Can you try this with a clean copy of the release version of AMBER 10? Amber
10 from the tar.gz release file and see if you see the same issues. Maybe
someone broke something inside the AMBER 10 tree.
All the best
Ross
From: Seetin, Matthew [mailto:Matthew_Seetin.URMC.Rochester.edu]
Sent: Thursday, August 13, 2009 1:44 PM
To: Ross Walker
Subject: RE: qmmm_div in amber10
Thanks for the feedback. It looks like the -nosanderidc option will fix my
issue in the short term, but I'm still not able to compile without that flag
set. To verify that I wasn't crazy, I did a completely clean checkout of
AMBER 10 from cvs, and I redownloaded all the latest bug fixes and applied
them all. If I try to compile without the -nosanderidc flag, the build will
halt with the following error:
ifort -c -w95 -vec_report0 -mp1 -O0 -FR -I../../sander -I../src/include
-DHAS_LAPACK -DQMMM -o ../obj/freq_qmmm.o ../src/freq/freq.F90
fortcom: Error: freq.F90, line 11: Name in only-list does not exist.
[QMMM_DIV]
use qmmm_module, only: qmmm_struct, qmmm_div
---------------------------------------^
fortcom: Error: freq.F90, line 38: The structure-name is invalid or is
missing.
dimension dcl(3,qmmm_div%ntotatm+qmmm_struct%nlink)
------------------^
fortcom: Error: freq.F90, line 38: A specification expression object must be
a dummy argument, a COMMON block object, or an object accessible through
host or use association [QMMM_DIV]
dimension dcl(3,qmmm_div%ntotatm+qmmm_struct%nlink)
------------------^
compilation aborted for ../src/freq/freq.F90 (code 1)
make[3]: *** [../obj/freq_qmmm.o] Error 1
make[3]: Leaving directory `/home/mseetin/amber10/src/dcqtp/mod'
make[2]: *** [libdivcon.a] Error 2
make[2]: Leaving directory `/home/mseetin/amber10/src/dcqtp'
make[1]: *** [divcon] Error 2
make[1]: Leaving directory `/home/mseetin/amber10/src/sander'
make: *** [serial] Error 2
It looks like the version of freq.F90 currently in the AMBER 10 cvs tree is
referring to changes made to qmmm_module.f in AMBER 11, but that weren't
made for AMBER 10.
Matt Seetin
Dave Mathews Lab
University of Rochester Medical Center
-----Original Message-----
From: Ross Walker [mailto:ross.rosswalker.co.uk]
Sent: Thu 8/13/2009 2:15 PM
To: Seetin, Matthew
Subject: RE: qmmm_div in amber10
Hi Matthew,
I am a little confused here. Is this AMBER 11 you are checking out that does
not work or Amber 10? Note AMBER 10 from CVS does not include bugfix.all.
You have to apply this.
Amber 10 should compile no problem since it was the release code. If you are
in AMBER 11 then a LOT has changed recently to the point where you should
probably delete you AMBER 11 tree and check out a brand new one.
Note in AMBER11 -nosanderidc is now the default so divcon should not get
compiled.
AMBER10 you can force this by adding the -nosanderidc flag to configure. I
recommend doing this.
Try this - with a make clean as well and let me know if you still get
problems.
All the best
Ross
From: Seetin, Matthew [mailto:Matthew_Seetin.URMC.Rochester.edu]
Sent: Thursday, August 13, 2009 9:51 AM
To: ross.rosswalker.co.uk
Subject: qmmm_div in amber10
One student in our lab was having some difficulties with AMBER 10 the other
day (don't worry, we're using 11, too, but with the recent difficulties, I
confess we've been using both), and I noticed he for some reason he was
using a pre-release version of 10 from who knows when back when we were
working on that rather than the released code. No problem, I just checked
out the current version for him and compiled it. Unfortunately, it wouldn't
compile. I got an error when compiling freq.F90 that it couldn't find
qmmm_div in the qmmm_module. Poking around in the cvs logs, it looks like
you created qmmm_div in 10/2008 to streamline some of the divcon stuff for
AMBER 11, and the change was reflected in both qmmm_module.f and freq.F90
for AMBER 11, but the change was also made in AMBER 10 in freq.F90 only.
qmmm_module.f in AMBER 10 hadn't been touched since 2/2008. Is this
something that you did deliberately so that there's something I need to
change on my end to get this to work? Do you have a recommendation for how
to get this to work?
Matt Seetin
Dave Mathews Lab
University of Rochester Medical Center
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Wed Aug 19 2009 - 22:34:33 PDT
