Please show the logs of it failing.
Are you running different firmware on the different hubs?
@bravenel I have been able to reproduce @mluck's issue on 2.4.0.151 (C8) but not on 2.3.9.200 (C7).
Steps to reproduce:
-
Create new hub variables of type number:
"Test - Hub variable 1" = 5
"Test - Hub variable 2" = 6
-
Create new RM 5.1 rule and name it.
-
Enable Required Expression and set it to Variable "Test - Hub variable 1" > 2
-
Create the trigger and set it to Variable "Test - Hub variable 2" > 20
-
Create a logging action just to have an action in the rule.
-
Set logging to Events, Triggers, Actions
-
Click Install Rule and then Done for good measure and then go back into the rule
Of note at this stage is that Event Subscriptions for the rule only have "Test - Hub variable 1" listed:
- Change the variable "Test - Hub variable 2" to a value larger than the trigger value (which is >20). In this case I set it to 258, which should trigger then rule since the required expression is already True. The Location Events log confirms that the variable was changed:
The two events at 19:02 and 19:03 are from when the variables were created.
Logs from the RM rule show nothing. The only events in the logs are from when I first clicked the "Install rule" button and then the "Done" button.
Still only the one variable in the event subscriptions of the rule is "Test - Hub variable 1"
I've tried setting the two variables to different values both under and over the conditions in the trigger and the Required Expression and hit "Update rule" and "Done" on the rule several times during this testing, but nothing has managed to get the second variable into the event subscriptions (which I assume is a requirement for the rule to trigger) or get the rule to run the actions.
1 Like
Yeah, there was a new bug introduced with a recent bug fix concerning variables in Required Expressions, and evidently it is not creating the right Event Subscriptions in this case. Fix will be forthcoming.
6 Likes
You really need to create a regression test ...
Yes, except for the fact that for such a test to be effective it would be several orders of magnitude larger than RM itself. Combinatorics cause such testing to rapidly expand to impossible dimensions. Attempts to guess where to trim or limit such tests leads to a meaningless outcome. So your suggestion is actually pointless.
Oh, and don't attempt to start a debate about this -- this is CS not EE!
What is a CS?
I am not attempting to start any debates. But it is what it is.
There are too many new bugs introduced with untested changes.
Something definitely must be done.
This is my last comment.
How to deal with all this things is your call.
1 Like
Computer Science.
Perfect
2 Likes
Yes, you and any other concerned user should join the Beta test program so that issues like these could be spotted before general release.
2 Likes
I respectfully disagree. The concept is perfectly correct in concept but in my experience often breaks down in practice.
Just one man’s opinion, but I find it refreshing how the Hubitat team openly acknowledge bugs when they surface and then they usually deliver a fix in days or sometimes even hours! And I’ll sheepishly admit that the last payment I made to HE was a year or two ago, so I align my expectations accordingly.
And I would hate it if Hubitat became like a typical software company—denying the existence of bugs, delivering fixes in cycles of months-to-years, and being completely opaque in the process. JMHO
5 Likes
This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.
