Talk: The View from Operations - Tasha Summers, NSLS-2 Operator

TS: Use NX to login from home.

Many of the early BOY screens have lots of rules or scripts (JS or Python) behind them, because of the need for speed in creating them at the time; better to create PVs on local Soft IOCs, will re-work in the future when time is available. Some of these pages have already been abandoned.

KS: Many of the problems with CSS may have fixes but they’re not fixed on a branch which can be used for stable releases. Stability is the highest priority. Different apps are at different stages of maturity in the development tree, agile development is the aim but not there yet.

GC: Every site that has their own release of CSS with their own apps has the same problems. Trying to have a common distribution of tools that work together.

TS: In many cases screens get copied from existing screens which work but are not well-engineered, maybe providing “best practices” examples would help.

AJ: Suggestion: CSS-BOY version of Perl::Critic module that looks at screens and gives indications of bad practices.

KK: Automatically run that and have it log the results on load?

-- Break --

Talk: User Experience of Channel Finder Service at NSLS-2 - Lingyun Yang (physicist)

Python notebooks used for commissioning and shown during talk at http://nbviewer.ipython.org/github/lingyunzjdt/aphla/tree/master/notebooks

KS: Experimented with using MongoDB instead of MySQL, from 4s => 120ms search time.

Talk: MongoDB and V4 based Archive Service at BNL - NM

Discussion tomorrow if time available.

Talk: Experimental Logging and Data Management Environment for NSLS-2 Beamlines - Arman Arkilic

AA: Showed the use of VisTools Python environment yesterday.

Using Python API to the Channel Archiver from Australian Light Source.

GS, KS, GC: discussion about advantages of switching to MongoDB, doesn’t work in some cases.

BD: No agreement on a common set of data to store across beamlines, which is why BNL went to a simple name/value pair schema in MongoDB.

-- break --

Talk: pvDS (PV Distributed Service) AKA Multicast - MS

(Talk was uploaded to the Tuesday AM hangout shared area.)

MD & KK questioned the apparently high packet-loss that MS measured in slide 4.

MS: Experimental development, will try to integrate C++ version with David’s AD plugins.

MD: Our use-case is only 2 destinations, why not just put a second network card in the camera machine?

BD: I want to know how far this can go.

Simon Ebner: PSI use 0MQ broker to do distribution of image data, either point-to-point or pub-sub. Saturate a 10GB network with this, the CPU is not the bottle-neck.

KK: SNS would also have very few (two) clients. One for saving Nexus data files, the other for display.

SE: High-rate & volume detectors need multiple streams anyway, but they hide that internally.

Talk: PVA py - AJ for SV

NOTE: order of fields in a structure?

NOTE: Why c.put(PvInt(7)) and not just c.put(7)?

DH: How to set an ID on a structure?

AI: Add support for it (if not already there).