- Friday, July, 2018 at 10 am
- CTA eZuce, no password. The connection details are here
- Collective minute taking here. Please help fill it during the call!
- Gammapy very slow progress in the past year, now development very active. See pulse in last month.
- Gammapy coding sprint last week at MPIK. See webpage.
- To try to keep the momentum (at least partially), we'll continue with weekly Gammapy calls Fridays 10 am.
- Note that CTA is also very active now. We need a clean and powerful Gammapy package, v1.0 and a paper this fall.
- Concerning cleanup / deprecations:
- We've been accumulating code for years; see here; the result was a very messy, unfriendly code base. Dozens of classes that basically aren't needed at all or are duplicated.
- Bad for users, but also I think one of the reasons development was glacially slow in the past year.
- We have to clean up and improve. So last week I deleted 10,000 lines of code / tests / docs, the parts that I thought were uncontroversial.
- Today we should discuss what changes go in v0.8 (hopefully in ~ 1 week) and which we postpone to v0.9 or even later.
- Note that v0.7 will always be as-is, you can use that. But of course, if you need to upgrade to v0.8 for an analysis, we want to support you. Either keep something or help you rewrite your script to the new code.
- What to do about the stuff in gammapy.scripts?
All added there with "not sure how to do this properly, not much time now, let's just put it there".
It was always clear that this has to be rewritten / relocated / merged with existing code.
Removal of
CTASpectrumObservationandCTAIrf(GH 1517) for v0.8 or v.9? Should we moveSensitivityEstimatorandSpectrumAnalysisIACTto gammapy.spectrum? - What about the SkyImage / SkyCube transition? A lot of the rewrite using gammapy.maps is done! The main remaining part are the IACT and Fermi-LAT image estimators. And the rewrite of the ring background estimation using maps is blocked by the presence of the IACTBasicImageEstimator. My suggestion would be to remove the FermiLATBasicImageEstimator now (see GH 1545). Then we update high-level docs and do v0.8 in ~a week. Finishing the transition, full removal of SkyImage / SkyCube / IACTBasicImageEstimator will then be in v0.9. Note that rewriting script from SkyImage / Skycube to use gammapy.maps is not hard. See intro_maps.ipynb. We're happy to help!
- Let's now move on to first 3D and DC1 analyses, and then to the dev hangout and open PRs and issues and the v0.8 milestone.
- Intro: see analysis_3d.ipynb and simulate_3d.ipynb
- Fabio: CTA DC1 - RX J1713 PSF issues here
- See v0.8 milestone
- See open pull requests
- What are the priorities to support CTA DC-1 analyses?
- Distribute developments for next week.