CmdStan setup

38 views
Skip to first unread message

Daniel Lee

unread,
Mar 11, 2014, 10:08:22 AM3/11/14
to stan...@googlegroups.com
It's not linked very well -- and I'm sure it's not clear (yet).


On Tue, Mar 11, 2014 at 10:01 AM, Bob Carpenter <ca...@alias-i.com> wrote:
I'm afraid I hate config so much that I haven't
even tried the new CmdStan setup.  Speaking of which,
are there instructions somewhere of what I need to
do other than emails?  

 
I still need to get the manual
refactored for the next release.

I've been meaning to start on that, but I haven't yet. I figure we should first move things into the CmdStan manual before pulling things out of the Stan manual.


By the way, for users, all they will do is download the tarball that we packaged (not the one that clones directly from github). That will have a copy of Stan inside and they won't have to do any git fiddling.


Daniel

Bob Carpenter

unread,
Mar 11, 2014, 10:42:41 AM3/11/14
to stan...@googlegroups.com
Thanks --- I can be the guinea pig.

- Bob
> --
> You received this message because you are subscribed to the Google Groups "stan development mailing list" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to stan-dev+u...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

Michael Betancourt

unread,
Mar 12, 2014, 7:08:55 AM3/12/14
to stan...@googlegroups.com
For development I found it easier to play with the submodule by hand.  So instead of running

make stan-update/feature/foo-bar

I did

cd stan
git remote set-url origin https://github.com/stan-dev/stan.git
git checkout develop

Will this mess up the submodule juju?  In particular I couldn't push without updating origin from git:// to https://.

Daniel Lee

unread,
Mar 12, 2014, 9:38:01 AM3/12/14
to stan...@googlegroups.com
If you're not going to work with CmdStan, I think that's fine. The idea of the submodule is to be able to keep everyone playing with CmdStan in synch with a commit of Stan.

I think what you've done is checked out Stan within CmdStan and not as a submodule.

When you change stan within cmdstan and you type 'git status' from cmdstan, do you see changes to the stan folder?

Michael Betancourt

unread,
Mar 12, 2014, 9:43:16 AM3/12/14
to stan...@googlegroups.com
When you change stan within cmdstan and you type 'git status' from cmdstan, do you see changes to the stan folder? 

13:41:38[~/Documents/Inference/stan-dev/cmdstan  (develop)]-betan: git status
# On branch develop
# Changes not staged for commit:
#   (use "git add <file>..." to update what will be committed)
#   (use "git checkout -- <file>..." to discard changes in working directory)
#   (commit or discard the untracked or modified content in submodules)
#
# modified:   stan (new commits, modified content)
#
no changes added to commit (use "git add" and/or "git commit -a")

It looks like it isolates the changes to repo.  Ideally I'd like to develop both cmdstan and stan from a single
directory so that I don't have to keep two working copies hanging around. 

Daniel Lee

unread,
Mar 12, 2014, 9:56:22 AM3/12/14
to stan...@googlegroups.com
It looks like I didn't set up the make targets correctly. I'll make it do what you're doing.


Daniel



--

Daniel Lee

unread,
Mar 12, 2014, 11:03:58 AM3/12/14
to stan...@googlegroups.com
Michael, thanks for helping me fix that.

If you check out from cmdstan again, it should work now. The fix was to a file called .gitmodules. If you look inside it now, you'll see I use https:// instead of git://

I had to wipe my cmdstan and then clone again for it to take place.


Daniel

Ben Goodrich

unread,
Mar 12, 2014, 12:35:14 PM3/12/14
to stan...@googlegroups.com
Also, is it possible to run Stan make targets from the CmdStan directory or do you have to drop down into the Stan subfolder to do that? Things on the Columbia cluster are working, but I have to rewrite the script that calls each test individually.

Ben

Daniel Lee

unread,
Mar 12, 2014, 12:36:55 PM3/12/14
to stan...@googlegroups.com
Ben, which targets are you referring to? I'll try to help where I can.

Models can and should be built from the CmdStan directory. The model tests are now in CmdStan and not in Stan.

Stan specific tests need to be run from the Stan subfolder.


On Wed, Mar 12, 2014 at 12:35 PM, Ben Goodrich <goodri...@gmail.com> wrote:
Also, is it possible to run Stan make targets from the CmdStan directory or do you have to drop down into the Stan subfolder to do that? Things on the Columbia cluster are working, but I have to rewrite the script that calls each test individually.

Ben

Ben Goodrich

unread,
Mar 12, 2014, 1:30:16 PM3/12/14
to stan...@googlegroups.com
Okay, that is what I thought.

Daniel Lee

unread,
Mar 12, 2014, 1:31:25 PM3/12/14
to stan...@googlegroups.com
Sorry -- did that mess up everything you were working on? Are you on a specific branch?

Ben Goodrich

unread,
Mar 12, 2014, 3:06:39 PM3/12/14
to stan...@googlegroups.com
It is fine, or at least it will be this week. Back when everything was under the stan repo I could do the equivalent of test-all but call it one test at a time so it could use one core per test. Now, it is just necessary to also do that for the tests in CmdStan.

Ben

Daniel Lee

unread,
Mar 12, 2014, 3:07:58 PM3/12/14
to stan...@googlegroups.com
Ok. If there's anything that could help, let me know. Like creating a make target that will print out all the tests individually.

Ben Goodrich

unread,
Mar 12, 2014, 4:06:01 PM3/12/14
to stan...@googlegroups.com
On Wednesday, March 12, 2014 3:07:58 PM UTC-4, Daniel Lee wrote:
Ok. If there's anything that could help, let me know. Like creating a make target that will print out all the tests individually.

That would help if it were formatted with one line per test so that I could easily redirect it to a file. Right now, I am just doing calls to find that mostly duplicate what is inside make/tests but it would be less brittle to have a make target for that purpose. The main thing is that, while I can pre-build all the common dependencies, each subsequent call to make should run exactly one test file.

Ben

Reply all
Reply to author
Forward
0 new messages