Miscellaneous fixes #842
Miscellaneous fixes #842
Conversation
These three add_op calls were all the same, so took it out of the if statement.
|
FYI I have no idea why TravisCI is borking on the builds for NumPy < 1.9. I made an issue with Conda to see if they have any insight. |
Yeah, they must have changed something on their end. The build also failed for #816 after you squashed it, and it wasn't failing before, so I'm quite sure it's not us. |
|
Yeah, this all looks good. |
|
LGTM too. |
Previously this was possible in the `Config` constructor, but not in `Config.configures`. This change makes it more consistent, and easier to configure multiple classes on an existing config.
This change can provide a legitimate speedup, since the function call happens once per timestep. The check should be pretty fast.
Similar to how they're done for NeuronTypes.
tbekolay
force-pushed
the
misc-fixes
branch
from
efb13b9
to
28704ec
Compare
|
OK, I amended that commit so it adds things to |
|
Looks good! I added a little fix to make the list of operators in Also, you reminded me that learning rule types should also be frozen objects. I forgot to include them in #758. |
|
If that fixup looks good to you, @tbekolay, I'll merge. |
|
Yep, looks good to me! |
Now, all Operators accept at `tag` argument which is part of the string representation. Also added a few sections to the `strings` notebook to demonstrate the Signal and Operators string representations, as well as some methods of comparing two sims by looking at the string representations of all of the operators. Also added some labels for unlabelled nodes in networks.
These are a small collection of things that I changed while working on the Frontiers paper. The only user facing change, I believe, is the
Config.configureschange, so it has a changelog entry associated with it.If any of these things are controversial, I'm happy to spin it into its own PR.