Public Lab Research note

Web Working Group Monthly Recap 7: September 2015

by warren | October 09, 2015 17:01 09 Oct 17:01 | #12284 | #12284

This month's work was a bit sporatic (for me, at least) because of the Public Lab staff retreat and the following week's desperate catch-up. Still, lots of work has continued on Spectral Workbench, and there've been a few other highlights as well.

Many apologies for the late update this month! I was just too overwhelmed with post-retreat work to get this out the door until now.

Systems administration

Sebastian Silva (@icarito), our sysadmin, has been hard at work tuning and optimizing our systems, and I've worked with him to push through a few fixes and changes, including:

  • a fix for callouts, which is where in comments you can use @username syntax to ensure someone gets an email notification of your comment (this had been broken for some time, unfortunately!)
  • a mix of small fixes for notification emails
  • a MapKnitter process optimization which was tested last night, and seems to work (please report any exporting trouble you have)
  • an upcoming mail server switch which we hope will resolve some of the issues with PL email going to some folks' spam inboxes


Sebastian has also put up the long-awaited 2.0, although there's a remaning bug where uploading doesn't actually work (so it's not useful yet!) but we hope to have that resolved shortly and will announce then.

Spectral Workbench 2.0

Finally, LOTS of work has gone into a new calibration system for SWB 2.0, as documented here, which has also led into a very interesting and productive discussion of calibration references, NIST, and possible errors in Wikipedia's published fluorescent bulb reference.


The calibration system and a new root-mean-square-error based assessement metric for calibrations are big features which promise to improve accuracy and precision for everyone, while relating our data more explicitly to well-known external references like NIST.

I've also begun to worry about how much longer this big upgrade is taking -- there've been plenty of distractions and urgent other work, but I also decided to move 12 issues into a 2.1 release, for some future date.

Still remaining in 2.0 and the graphing overhaul milestones are 44 issues. Gosh, I hope I can get most of those done by the end of October, but I don't know.

Next up will be automated JavaScript testing -- this'll help speed development since I'll be able to tell if JavaScript changes are breaking any of the more complex systems which now exist only on the client side.

In all, 32 issues were closed since Sept 1 -- a full list is as follows:

Thanks again to our contributors, both coders and non-coders -- and if you know any coders who might like to get involved, point the at!


Login to comment.