New release? about dill HOT 10 CLOSED

I'm not sure that dill 0.1 is broken, but it is a few years old.
When you pip install, you get the "0.1" version...? Hmm.

I am working on a new release. However, it's a few months late... http://trac.mystic.cacr.caltech.edu/project/pathos/roadmap.

Every couple months, I do post "dev" releases here: http://dev.danse.us/packages/

from dill.

I couldn't get it to work.

One idea would be to do a dill 0.1.1 release then which is the current dev release. It can only be more stable than the current 0.1.

from dill.

The release of dill is so old that pip didn't exist then, so I think the issue is that it assumes setuptools. I'll address this next week... if that's the case, I'll just start cutting rc's for the pending release with the fix for pip.

from dill.

I see. Just found a fresh box and installed dill with pip. So if you have pip-1.3, the latest version of dill you can get is the "stable" version dill-0.1 because pip assumes that you will only ever want the stable versions installed.
However, if you have pip-1.4, you can use the "--pre" flag to install the latest development release... and then you get the most recent dev release version of dill... and it installs no problem.

Since I have a new stable release pending, and there's an existing way to get the most recent dev release with pip... I'm don't see that I need to do anything here. The new stable release should be out soon, and then it won't matter if you have pip-1.4 or pip-1.3.

from dill.

Sounds good, thanks for looking into this.

from dill.

@twiecki: posted new release to pypi

from dill.

Thanks, @mmckerns !

from dill.

@twiecki: Np. also note, it's 3.x compatible, which is nice... and opens up a lot of new work to do.

I may want to talk to you about parallel/distributed MC at some point. This year I have to build such a code, and am considering leveraging/extending/forking PyMC.

from dill.

@mmckerns Sounds interesting! Definitely drop me a line at thomas.wiecki at gmail.com. You might also be interested in pymc3 (now on the master branch at https://github.com/pymc-devs/pymc) which is a complete rewrite and already has parallel sampling capabilities (but no truly distributed samplers yet).

from dill.

@twiecki: Cool, I just saw that. I will definitely send you an email.

from dill.