Not sure how the formatting will come out on this (looks fine to me) but
here are the notes I took from the sprint planning meeting.NuPIC 10/25/13
TimestampNote3:00PMAttendance: Matt Taylor, Subutai, Austin, Scott, Matt K,
Fergal, one or two others I can't see+3:00PMMatt: Using huboard.com to help
with Github issues+3:01PMMatt: Still working on labels. Requires
permissions to edit so let Matt know if you have input.+3:02PMMatt:
Milestones == sprints. JIRA will be left around but no longer being used
and will attempt to make read-only.+3:03PMGreen icons let you see if issues
are true issues or PRs since pull requests show up in the issues list.+
3:04PMPR and issue numbers are unique so there is no ambiguity when you do
a #232 reference.+3:06PMCMake - unsure of status, David to fix CI issue.
Can hopefully merge without affecting the current Make pipeline.+3:08PM#294
- removes Python dependencies from NuPIC. Relies on user to install deps.
requirements.txt included to make this easy. nupic- also available as a
base with the dependencies installed. C++ dependencies to come
later.+3:10PM#302
- Hung up on operator<< issue. Might have to use a temporary non-template
solution. Scott and Subutai to follow up.+3:11PM#318 - Matt not sure how to
handle it. Assigned to Scott to determine what to do.+3:11PMScott to check
to make sure we are still running test that compares TPs in NuPIC CI+3:15PM#321
- Unclear if new features (auto, etc) are backwards compatible with
compilers that people are currently using. Assigning to Scott (but in
backlog) to check if it works and follow up on the issue or message mailing
list to see if it is an issue for anyone else.+3:16PM#325 - To be merged,
keeping assigned to Scott.+3:17PM#326 - Remove external deps from NuPIC
repo - assigned to Austin. Python portion close to being merged. Keeping
this to continue tracking for C++/binary dependencies.+3:18PM#333 - Putting
in the backlog for now. Scott: seems like people on mailing list want this
but not a lot of opinions on how to implement. Unclear when this will be
tackled (it is a big task).+3:21PMDiscussion on huboard. Link is
huboard.comand requires giving Github public access to the site.
Fergal suggest
sending a mailing list message about issue tracking changes since many
people don't follow wiki changes.+3:22PMSprint is pretty light since
Grok-ers are pretty tied up with product work. Some issues can be tackled
at hackathon.+3:25PMSubutai: We do plan to test new spatial pooler
implementations and move to Gil's new C++ SP as the default. Creating an
issue to track progress on this.+3:28PMMatt is trying to find a good place
for teams to collaborate. challengepost.com is being considered. If all
else fails, Matt to use Google Drive. Fergal is going to check
challengepost to see how it looks and will get back to Matt.+3:28PMMatt: We
have hackathon shirts!+3:29PMMatt: Haven't had time to update roadmap.
Hackathon schedule up on events page but not solidified yet. Hard schedule
by Monday. Everything is optional.+3:30PMFergal: 1. Soft tip - regarding
Vagrant VM - downloaded latest NuPIC code and tried it on an old VM and it
was a bit of a mess. Advises that we create some functionality for
synchronizing the VM to the latest (update bashrc, etc).+3:30PMSubutai:
There is an issue for that.+3:32PM#341 created to track VM
updates+3:35PMFergal:
2. Would like time on schedule - 10-15 minutes is fine - for presentation
on discussions he has been having with machine learning people. Wants to
help coalesce all of the discussions on ML and elsewhere. Fergal to send
description to Matt. Talk inspired by talks about hierarchy and Geof Hinton
discussions.+3:35PMSubutai - there is a session on comparing NuPIC
algorithms with other approaches+3:38PMMatt Keith - Working with building
on stock Mavericks (clang), etc and will follow up when he makes progress.
Current issue is with Python not finding libyaml. Has C++ code compiling
though.+3:39PMHackathon is at Tagged in San Francisco. You can still RSVP,
it is a free event. Getting close to our limit of 75 but it is a soft limit
and we can increase if there is interest.+3:40PMFergal: Can we live cast
the hackathon? Matt: Yes. (We streaming it last time and plan to do it this
time as well.) Fergal: Likely will have many more remote observers this
time. Matt: Happy to have remote participants demo on the big
screen.+3:45PMScott:
Jeff is sending an email about his idea for a new scalar encoder. Hoping
someone from the community will pick it up. Scott is happy to point out
example code and answer questions for implementing it. Pretty
straight-forward task and quite elegant.+3:45PMEnd+
_______________________________________________
nupic mailing list
[email protected]
http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org

Reply via email to