Closed stuwilkins closed 1 year ago
All of the above
OK. If this is disruptive to user operations and you lack the bandwidth to address it in time, I'm willing to help.
@danielballan don't worry, I think it is good that I (we) have to change this. I will work on this from home wed/thurs to get all inline. I may need help with handlers etc? Where is the idea of adding a config file for defining how the handlers get assigned to specs? That is by biggest need at the moment in this respect (CSXTOOLS)
Oh and we do need to get FastCCD data into databroker correctly....
Where is the idea of adding a config file for defining how the handlers get assigned to specs?
Nominally implemented and deployed, but when I looked up the relevant code to back up that claim, I found that it's undocumented, untested, and broken. All those problems are addressed in https://github.com/NSLS-II/databroker/pull/305, which will be merged and deployed at CSX today.
Oh and we do need to get FastCCD data into databroker correctly....
With regard to the timing of when documents are created?
@stuwilkins: @danielballan don't worry, I think it is good that I (we) have to change this. I will work on this from home wed/thurs to get all inline.
csxtools still has a filestore import, which made it un-importable in the 2017-3 conda environment. It would be nice to fix it in time for 2018-1. Do you need help with that?
attn @wen-hu
@stuwilkins I looked through all the branches on your fork, and I don't see one that addresses the subject of this issue. @ambarb mentioned there might be one on srv3. Any guidance?
unsure if this particular instance was ever fixed / addressed. @stuwilkins let me know if the changes below do not address the issue that started this thread
Taking over the issue for the current problems. Using get_table()
because .data()
does not work with current version of databroker
at CSX.
Noting that we used databroker 0.13.0
doesn't play nice with the default HDF5 handler for area-detector. Only get_table()
seems to work (which is why CSXtools still uses it. Testing with @danielballan confirmed handler issue. However, the issue is not present in databroker 1.0.2
@leofang FYI, but I don't think we are ready to make the switch yet, just thought you would be interested.
stale
addressed also with https://github.com/NSLS-II-CSX/csxtools/pull/86
As far as you know, is csxtools broken by our API changes, invoking our deprecations, or none of the above?