{chem,math,lib}[intel/2017b] NAMD 2.12, MDTraj 1.9.1, molmod 1.4.3, yaff 1.4.2 by tovrstra · Pull Request #5317 · easybuilders/easybuild-easyconfigs

tovrstra · 2017-11-09T10:48:43Z

I have some questions about the NAMD easyconfig:

The download from the NAMD website (requires login) is a tar file and not a tar.gz file. Still, the .gz extension is present. I've added a comment to reduce the confusion. I hope this is OK.
Is there a way to change the easyconfig such that it uses Tcl/8.6.7-GCCcore-6.4.0 instead of Tcl/8.6.6-intel-2017b. The former is loaded with Python/3.6.3-intel-2017b and may cause conflicts when loading NAMD afterwards. How should we deal with this?

boegel · 2017-11-09T11:22:13Z

@tovrstra Which comment are you referring to for NAMD, I don't see it in the easyconfig file?

If the source filename is incorrectly named, you probably need to specify the correct extraction command yourself, since EasyBuild relies on the file extension being correct. See http://easybuild.readthedocs.io/en/latest/Writing_easyconfig_files.html#common-easyconfig-param-sources-alt.

To fix the Tcl issue, why not bump the Tcl version for NAMD to 8.6.7? Note that Tcl-8.6.7-GCCcore-6.4.0.eb suffices to resolve that dependency; since EasyBuild v3.x, eb will also consider subtoolchains if an easyconfig with the 'parent' toolchain (intel here) is not found.

boegel · 2017-11-09T11:23:09Z

+# https://github.com/mdtraj/mdtraj/blob/master/devtools/conda-recipe/meta.yaml
+sanity_check_paths = {
+    'files': [],
+    'dirs': ['lib/python%(pyshortver)s/site-packages/%(namelower)s-%(version)s-py%(pyshortver)s-linux-x86_64.egg/%(namelower)s'],


This line is too long, so it should be trimmed somehow (that's also why the tests are failing in Travis, we're stricter w.r.t. style now in easyconfigs).

You can just check for lib/python%(pyshortver)s/site-packages here, since EasyBuild already does an import mdtraj check.

Good. How does easybuild now that it should try import mdtraj? The module name is not necessarily related to the package name.

By default is uses name.lower(), you can override it with:

options = {'modulename': 'foo'}

Allright! Thanks. In that case, I can remove this line from the yaff eb file.

boegel · 2017-11-09T11:24:07Z

+toolchain = {'name': 'intel', 'version': '2017b'}
+toolchainopts = {'pic': True, 'usempi': True}
+
+sources = ['NAMD_%(version)s_Source.tar.gz']


@tovrstra Please include a SHA256 checksum here too (you can use --inject-checksums for that).

boegel · 2017-11-09T11:25:07Z

+
+source_urls = ['https://github.com/molmod/yaff/releases/download/%(version)s']
+sources = [SOURCE_TAR_GZ]
+patches = ['yaff-1.4.2-add-missing-include.patch']


Please include SHA256 checksums for both sources and patches (pro tip: use --inject-checksums).

Also avoid hardcoding the yaff version in patches, use %(version)s instead.

boegel · 2017-11-09T11:59:34Z

Hmm, seems like I have been modifying the downloaded sources myself (i.e. by zipping it again) up until now, because the source file I have is gzipped. It's not when I redownloaded it though.

Does this ring a bell for you @akesandgren?

I just subscribed to the NAMD mailinglist and posted a question for this. I'll post the replies I'll get here.

The file I have on disk is NAMD_2.12_Source.tar.gz
But i don't remember having done anything to it.
And the older files i have are also tar.gz

Yes, the extension is tar.gz, but can you run the file command on it? When I do this, I get

$ file NAMD_2.12_Source.tar.gz NAMD_2.12_Source.tar.gz: POSIX tar archive (GNU)

So not gzipped, despite the extension.

@akesandgren Indeed, the downloaded file has .tar.gz extension, but it's really just a (non-gzipped) tarball.

I see the same thing when re-downloading it, so I suspect I just fixed that myself manually by gzipping it after downloading.

This approach here is better, but then we should also make it consistent (i.e. apply the same change, including a checksum, in the existing NAMD 2.12 easyconfigs).

... which is exactly what @tovrstra did, thanks. :)

You can follow up my message on the mailing list here: http://www.ks.uiuc.edu/Research/namd/mailing_list/namd-l.2016-2017/2442.html
No replies yet. Nobody seems to care, so don't expect this to change anytime soon.

I've looked around a bit: you can also get the source code here apparently: https://github.com/divijverma/NAMD/releases I'm not sure how legal and reliable that is. At least you'll get a .tar.gz when you're told so :)

We should stick to the official sources, and just deal with the faulty naming, it's an easy/clean fix now in the easyconfig.

tovrstra · 2017-11-09T12:46:54Z

It seems that Tcl-8.6.7 cannot be used as a dependency in NAMD-2.12-intel-2017a-mpi.eb? Why doesn't it use Tcl-8.6.7-GCCcore-6.4.0 as a dependency in that case?

boegel · 2017-11-09T13:04:54Z

@tovrstra Don't bump dependency versions in existing easyconfigs, since that may cause problems elsewhere (e.g. version conflicts on Tcl when this NAMD is reinstalled and then combined with existing modules).

Also, there's no easyconfig file for Tcl 8.6.7 that is compatible with inte/2017a.

boegel · 2017-11-09T13:05:24Z

Hmm, this worked fine without this patch though?

I really have no idea. I also don't understand why it passed the unit tests without setting lowopt=True. I'll check.

This is weird! Indeed, the patch is not needed for yaff-1.4.2-intel-2017b-Python-2.7.14.eb but it is still needed to let the tests pass for yaff-1.4.2-intel-2017b-Python-3.6.3.eb. Even stranger, the tests run fine on Delcatty without lowopt=True, while on Golett lowopt=True is needed. I'll revert the changes in yaff-1.4.2-intel-2017b-Python-2.7.14.eb but I can't make much sense out of the current behavior.

@tovrstra Delcatty is Intel Sandy Bridge, Golett is Intel Haswell, which may explain why lowopt=True is only needed for the latter for yaff.

boegel · 2017-11-09T13:06:36Z

@tovrstra For intel/2017a, GCCcore/6.3.0 is a subtoolchain, so Tcl-8.6.7-GCCcore-6.4.0.eb wouldn't satisfy that dependency.

But you shouldn't be updating dep versions in existing easyconfigs anyway... :)

tovrstra · 2017-11-09T14:01:12Z

ok. I'll fix that. Sorry...

tovrstra · 2017-11-09T14:04:03Z

I've removed the dependency change. It should be OK now.

tovrstra · 2017-11-09T15:03:29Z

Changes in yaff-1.4.2-intel-2017b-Python-2.7.14.eb were reverted.

boegel · 2017-11-11T14:24:14Z

+ thermochemistry and reaction kinetics. It uses a Hessian computation from a
+ standard computational chemistry program as its input."""
+
+toolchain = {'name': 'intel', 'version': '2016a'}


Hmm, this is the only one using intel/2016a rather than intel/2017b, does it belong to the same batch of software?

No, this is an older one, in fact unrelated to the other ones. I haven't converted TAMkin to work with Python 3 yet.

tovrstra · 2017-11-13T08:43:59Z

Are there any other changes needed. I think all is fixed now?

boegel · 2017-11-15T08:24:34Z

@tovrstra The TAMkin easyconfig is proving to be a bit of a PITA, since the tests seem to be hanging for some reason.

Maybe it's better to remove the TAMkin easyconfig here, and open a separate pull request for it to avoid it blocking this PR?

tovrstra · 2017-11-15T11:26:29Z

The TAMkin easyconfig was moved to separate PR: #5333

boegel · 2017-11-15T12:25:08Z

Test report by @boegel
SUCCESS
Build succeeded for 7 out of 7 (7 easyconfigs in this PR)
node2432.golett.os - Linux centos linux 7.4.1708, Intel(R) Xeon(R) CPU E5-2680 v3 @ 2.50GHz, Python 2.7.5
See https://gist.github.com/a6a14de5be0d025ed4d0726f6ccf715f for a full test report.

boegel · 2017-11-15T13:20:59Z

Test report by @boegel
SUCCESS
Build succeeded for 7 out of 7 (7 easyconfigs in this PR)
node2101.delcatty.os - Linux centos linux 7.4.1708, Intel(R) Xeon(R) CPU E5-2670 0 @ 2.60GHz, Python 2.7.5
See https://gist.github.com/6f8a482ae6d353d2735a2ba872c09656 for a full test report.

boegel · 2017-11-15T13:31:19Z

Going in, thanks @tovrstra!

Software needed for thesis Sander

088db3d

boegel requested changes Nov 9, 2017

View reviewed changes

boegel reviewed Nov 9, 2017

View reviewed changes

boegel requested changes Nov 9, 2017

View reviewed changes

tovrstra added 3 commits November 9, 2017 15:02

Fixups for NAMD easyconfig, also in other 2.12 configs

262b54e

Fix too long lines and missing checksums

f46d1a9

New Tcl config no longer needed

4d529f3

tovrstra force-pushed the cmmsoft5 branch from 409ab01 to 8eaa5f6 Compare November 9, 2017 14:03

Remove redundant lines from yaff eb

25e8b04

tovrstra force-pushed the cmmsoft5 branch from 8eaa5f6 to 25e8b04 Compare November 9, 2017 15:02

easybuilders deleted a comment from boegelbot Nov 11, 2017

boegel reviewed Nov 11, 2017

View reviewed changes

boegel changed the title ~~Software needed for thesis Sander~~ {chem,math,lib}[intel/2017b] NAMD 2.12, MDTraj 1.9.1, molmod 1.4.3, TAMkin 1.4.2, yaff 1.4.2 Nov 11, 2017

boegel added this to the 3.5.0 milestone Nov 11, 2017

boegel added new update labels Nov 11, 2017

Remove TAMkin Easyconfig for now

d5bec82

tovrstra mentioned this pull request Nov 15, 2017

{chem}[intel/2017b] TAMkin 1.4.2 #5333

Merged

tovrstra changed the title ~~{chem,math,lib}[intel/2017b] NAMD 2.12, MDTraj 1.9.1, molmod 1.4.3, TAMkin 1.4.2, yaff 1.4.2~~ {chem,math,lib}[intel/2017b] NAMD 2.12, MDTraj 1.9.1, molmod 1.4.3, yaff 1.4.2 Nov 15, 2017

boegel approved these changes Nov 15, 2017

View reviewed changes

boegel merged commit 8543b77 into easybuilders:develop Nov 15, 2017

Conversation

tovrstra commented Nov 9, 2017

Uh oh!

boegel commented Nov 9, 2017

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

tovrstra commented Nov 9, 2017

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

boegel commented Nov 9, 2017

Uh oh!

tovrstra commented Nov 9, 2017

Uh oh!

tovrstra commented Nov 9, 2017

Uh oh!

tovrstra commented Nov 9, 2017

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

tovrstra commented Nov 13, 2017

Uh oh!

boegel commented Nov 15, 2017

Uh oh!

tovrstra commented Nov 15, 2017

Uh oh!

boegel commented Nov 15, 2017

Uh oh!

boegel commented Nov 15, 2017

Uh oh!

boegel commented Nov 15, 2017

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants