Revisit how BIN_MIN and BIN_MAX are specified in das2stream

[jbfaden]

Chris asked about the format for das2streams with error bars and bounaries. Autoplot can export a dataset with min and max to a das2stream using code Chris wrote:

das2java/dasCore/src/org/das2/client/QDataSetStreamHandler.java

Line 287 in db034ea

if ( checkOperation.equals("BIN_MAX") ) {

Autoplot wasn't picking this up in its reader (it looks like there was a branch which would handle this but it was turned off and threw an error when I tried it). I've made it so that the branch it was taking looks for the .min and .max extentions. (mydata.max is the maximum observed value in the interval, mydata.min is the minimum.)

Note too that Autoplot doesn't show error bars by default. They are detected, though, and the scientist can enable them with the checkbox on the style tab.

Chris also asks that stddev be added, and these will be handled the same way. Though these look different in the QDataSet data model, Autoplot (and Das2) draws error bars and extremes the same way.

[jbfaden]

I've added stddev, and it looks like Autoplot automatically turns on the error bars in this case. Look for the next release. Also I'll add tests so this doesn't revert again. (I saw some comments by me about changing behavior in 2021, so clearly we need to lock in this functionality.

[jbfaden]

See https://jfaden.net/~jbf/autoplot/data/d2s/errorbars.stddev.d2s and https://jfaden.net/~jbf/autoplot/data/d2s/errorbars.minmax.d2s, and
https://cottagesystems.com/jenkins/job/autoplot-test148

[jbfaden]

Actually I'm going to reopen this because there's an old d2s parser and new one which goes straight to QDataSet, not via the old data model. I thought this new one was in production use, but apparently that's not the case. I'm going to fix this as well so there isn't a surprise in the future, and we should work towards using the new parser.

[jbfaden]

The new parser also handles this properly.

[cpiker]

Nice to go straight to QDataSet and skip the middleman. Do you plan to do any dead code removal on the old parser to prevent confusion, or keep it in main for quick reference?

[jbfaden]

There must be some reason why we/I switched back to the old one, but yes I think we would want to get rid of the code. It's also the only part of the system that uses the old data model so it would be good to get rid of it for that reason too. I'll start a ticket for that, too, but honestly I don't know who would be interested in funding that effort.