Automation Professionals' Integration Toolkit Module

3rd Party Modules
, , ,
242

Hey @pturmel,

I'm trying to figure out how I can calculate the time-weighted rate of change from a dataset with "t_stamp" and "value" columns with your toolkit expression functions, but i'm a bit lost with the starting at i=1 and ending at i(n-1) but still using i(n)

image
image735×146 6.87 KB

sample dataset:

"#NAMES"
"t_stamp","value"
"#TYPES"
"date","F"
"#ROWS","6"
"2025-04-11 14:00:00.000","10.0"
"2025-04-11 14:01:00.000","20.0"
"2025-04-11 14:02:00.000","30.0"
"2025-04-11 14:03:00.000","50.0"
"2025-04-11 14:10:00.000","100.0"
"2025-04-11 14:15:00.000","100.0"

Without adding "now" it should return "90" edit: if it wasn't working properly. If it was working properly it would return "6"

243

What a terrible example. That equation simplifies to (vn - v1) / (tn - t1). No need for the intermediate rows. (It does not yield 90, fwiw.)

The lag() function yields xr-1, with a null for the first row. Wrap lag() in a coalesce() to substitute in that first row. Or include an extra row as first, then cut it off with a where().

1 Like
244

Whoops, I had a rough day :sweat_smile: I think my wife might have cooked her car engine, my son got nits from school, and I obviously forgot how numbers work :face_with_crossed_out_eyes:

Oops, I typed the wrong value, should have been 6!

Correct, yes!

OK Ill give this a go, cheers!

1 Like
245

Try this:

transform(
	{path.to.source.dataset},
	sum(
		forEach(
			value(),
			if(
				isNull(lag()),
				0.0,
				(it()[1] - lag()[1]) * (toMillis(it()[0]) - toMillis(lag()[0]))
			)
		)
	) / (toMillis(value()[len(value()) -1 , 0]) - toMillis(value()[0,0]))
)

{Untested.}

246

I think we're both overcooking it:

transform(
	{this.custom.key},
	(value()[len(value())-1, 1] - value()[0, 1]) / 
	(millisBetween(value()[0, 0], value()[len(value())-1, 0])/1000/60)
)
247

Hah! Indeed.

Tuck away what I show, though: if you replace the denominator with a time unit constant expressed in millis, the expression yields the integral of v dt in that timebase.

2 Likes
248

:frowning:

IGN-8204, someday, will let you do negative subscripts on ordered sequences.

Maybe I can convince someone to let me do a shotgun pass to the expression system...

3 Likes
249

Ooooo! I should add that to the qvAt() function.

250

That was easy. :grin:

Automation Professionals is pleased to announce a new Production release of this Integration Toolkit. New features and changes since v2.1.0:

  • Tag Actors
  • tags() refactoring to return good quality even when some tag values are not good.
  • New qvAt() function to make quality of list and dataset nested values accessible, with python-like negative index support.

For Ignition v8.1: v2.1.1.251011523

4 Likes
251

...huh, is the expression system not running through jython -- is there a custom lexer/parser under there?

252

Expressions are pure java, except where a function author deliberately calls out to jython. In IA's stuff, only the runScript() expression hands off to jython.

In my toolkit, objectScript() is souped-up version of runScript(), and view() uses jython, too.

2 Likes
253

The reason users are advised to use expressions instead of scripts where possible is because the jython interpreter has significant overhead. And it is an interpreter, with all that implies.

254

Yep. It's generated code, but pure Java.

255

LR(1) lookahead, i'm gathering from the docs :slight_smile:
(i don't suppose y'all have a BNF for the expression language)

256

is there a set() (specifically, the list deduplication “feature") equivalent? (before I try to derive a creative one)

257

You could do something funky like where({array}, IndexOf({array}, it()) = idx()).
Seems to work in perspective at the least.

Pre-Expression:

[
  {
    "key": "value",
    "key_1": 2
  },
  "value2",
  {
    "key": "value",
    "key_1": 3
  },
  "value",
  {
    "key": "value",
    "key_1": 3
  },
  "value"
]

Post expression:

[
  {
    "key_1": 2,
    "key": "value"
  },
  "value2",
  {
    "key_1": 3,
    "key": "value"
  },
  "value"
]
1 Like
258

Just curious, what's the use case you have for this?

259

deriving a single working tagpath from “fuzzy matching” on unordered/unspecified path parts passed in to a view. could be from flex repeater instance parameters, implicit rowData parameters, cell value parameter, etc. Each method has limitations and I don’t want to hardcode fullpaths, but could end up with duplicates. I could just iterate over the duplicates, but thought might as well deduplicate them

speaking of - I’m sure I’ve figured out why folder tag bindings return ‘2’, but can’t remember at the moment.

for instance each highlight is the exact same embedded view, but

  • subview components are using full tag paths (results of system.tag.query from each row’s hidden fullpath column
  • most table columns are using derived paths based on cell value, rowdata and column viewparams with an “altPaths” list of possible subpaths/tagnames
  • % Valid column is using a completely different path that can’t be derived entirely from rowdata so using string replacement(indirection) on a “parameterized” “full path” coming from the same column viewparams “altPaths” list

CleanShot 2025-04-11 at 15.42.34
CleanShot 2025-04-11 at 15.42.34985×427 69.7 KB

260

Yes, you can emulate set() with something like this:

forEach(
	asMap( // Keys of a map are forced to be unique.
		forEach( // Convert simple list to list of pairs where the 2nd value is a dummy
			{path.to.list.of.strings},
			it(),
			""
		)
	),
	it()[0] // Extract the unique keys back out of the map.
)
2 Likes
261

Phil,

Just wanted to say that this module has once again saved me a bunch of work.

Needed to filter a dataset, because it couldn't be done in a Named Query, and I really didn't want to have to script the query, just be cause NamedQueries can not utilize a list to facilitate an IN clause.

where() to the rescue.

Sometimes, I wonder how much time I would waste without this module.

6 Likes