Dashboard 2.0 bugs I've found

Dashboard 2.0 is too unreliable. I can`t understand why good working Dashboard 1.0 is deprecated. Does somebody did regression tests before forget Dashboard 1.0?

Bugs what I found -

  1. Unable to play sound from node-red editor
  2. Unreliable event node, sometimes it works, sometime doesn't
  3. Too many errors -
  • iframe errors
  • errors every time when changing UI parameters via ui-control node (pic below)
  • some freezes 5-10 seconds (I think because of errors in browser)

I'm really thinking about return to 'deprecated' Dashboard 1.0 in my projects. I think moving to Dashboard 2.0 was a mistake.

What other bugs do you found in Dashboard 2.0?

Also it's impossible to fully hide sidebar. In Fully Kiosk Browser(Webview) sidebar is active when I set it to Always Hide. It appears with swipe from left to right.

Because Dashboard 1 is based on AngularJS v1 which stopped being updated in April 2022 and so is over 3 years out of date and may contain an unknown number of security and other issues.

If Dashboard 2 isn't meeting your needs, perhaps also consider whether starting to migrate to UIBUILDER might be an option.

The uncaught errors are super annoying as you can’t really debug them as everything freezes up, it’s the reason why I don’t//can’t use the dashboard

I don't see such problems.

Can you post example flows showing the bugs you describe please.

I am certainly not aware of any 'uncaught error' bugs.

The dashboard is certainly still missing some important features, but most bugs are fixed.

add a template node, enforce an error.

I didn't realize that you meant that the dashboard may not catch errors in the code that the user writes. I am not sure that would be possible.
Did D1 catch errors in scripts in templates?

Do humans make mistakes? Yes. Should the dashboard become unusable at that point?

Did D1 catch errors in scripts in templates?

Right, AngularJS v1 is discontinued and AngularJS v2 doesnt have backward compatibility. Now I understand reason appearing Dashboard 2.0.

I think UIBUILDER is great tool, but I'm not too familiar with programming languages for web develoipement.

Thanks :smiley:

Well, UIBUILDER may possibly be the easiest way to get started. Clearly you already have some knowledge since you are using ui_template nodes which require a little knowledge at least.

UIBUILDER gives you a working template right out of the box. It gives you some simple tools to turn data into basic visual elements in the browser. Even complex UI's often require very little HTML and often no JavaScript at all (data from Node-RED being connected direct to the HTML).

There are also many web components (a vanilla HTML standard) that can be used to assist with your pages including some that I've created that work closely with UIBUILDER/Node-RED. Again, these simplify the coding of your HTML.

Not pushing this on you, simply highlighting that, while UIBUILDER may seem somewhat complex from the outside, the reality is often seen to be different once you get started.

And, since you can use iframes to include UIBUILDER pages in Dashboard 1/2 (or the other way around), it is also possible to take it one step at a time.

My general advice to all is that if you find yourself using more and more complex ui_template's and CSS overrides with D1 or D2, and find yourself fighting with their frameworks, consider whether things might actually be simpler with UIBUILDER.

It might be possible to catch errors withing existing AngularJS/Vue managed data and methods but no, probably not possible to catch all errors in custom ui_template code. That must surely be up to the author of the template.

From the original post:

  1. Sound is going to be quite browser specific I think? Will depend on what browser API's are being used. I don't think D1 or D2 have sound-specific nodes do they? If not, then the ui_template must handle the errors.
  2. This one needs more info to understand.
  3. I don't think that it is really possible to understand the errors from the image without knowing more about the code. It is possible that all those errors could stem from a single error - without the code it is probably impossible to know.

Its one of the great NR features that GUI and backbone is independent. I was running with Dashboard 1 version from 5 years ago as this was the last one to support my old recycled tablets.

Maybe not „secure“ but that was never an issue for me.

I am now looking for a specific dashboard for eInk displays…

Hi @haegar33,

I think this needs to be raised in a new thread? Happy to help there

I'm working with embedded systems, and result of every string in my code I'm fully understand. Modern web programming seems like blackmagic for me. For me it's like huge onion where abstraction layers layes on another abstraction layers. As a resault, nobody can explain root of an error if it happens. It's hard to understand for me, thats why I'm not so familiar with modern web programming.... But it's modern reality, and I'm understand it.

For what it’s worth, I tried moving over to Dashboard 2.0 but gave up.

I mean no negativity whatsoever to the developers of Dashboard 2.0 and very much appreciate their efforts but just found it too glitchy / buggy. I still have little annoyances like icons completely misaligned in admin sidebar, various limitations with specific dashboard widgets, and to my mind not quite ready for production. I appreciate it’s improving all the time but I’m one of those people who needs something to “just work” which I find Dashboard 1.0 does.

Sure Dashboard 1 has annoyances too, namely it lacks responsive layouts, but much more predictable in the way that it works and therefore I still plough hours of time into making widgets and templates for my automations in Dashboard 1.0. Futile, I know, but it works for me, right now.

I also wish I could get over the mental block / inertia of learning how to use UIBuilder. I’ve no doubt it would fit my needs. I keep meaning to try harder. I think for me the idea of a blank slate scares me a bit, whereas Dashboard 1 solves this by just letting me whack a few gauges or text inputs on the screen to get me up and running, then I move to much more detailed UIs using template nodes and CSS.

Perhaps I may revisit UIBuilder with the help of AI to get me up and running…

Sorry if this post just adds to the “noise” but maybe it’s useful for discussion in some small way?

:smiley: Because it is!

A fair assessment.

Ah, well here we can differ. Actually, it is generally fairly straight-forward to debug. Not always, its true, but mostly. Of course, it helps if JavaScript library authors build in sufficient error handling and logging. Hopefully you will find that UIBUILDER is pretty good on that front. Not least because, I am not a "professional developer" and UIBUILDER is now around 15,000 lines of code (that covers both the Node-RED side which is by far the larger, and the client side which is much simpler). So no way I can always remember things. Logging is critical. And UIBUILDER has it in spades.

I get it, I've been messing on and off with web development on a personal front since the 1990's and I absolutely don't pretend to be an expert.

HOWEVER, that is the premise on which I built UIBUILDER in the first place! I try to deal with the messy parts so you don't have to.

When using UIBUILDER, the amount of any code you need to write is minimised without locking you into one specific way of working or a specific framework. This is really important to me. Frameworks come and go, but vanilla HTML, CSS and JavaScript are unlikely to ever go away and, even more critically, they do their darndest to remain backwards compatible - unlike many (most?) frameworks. This is also true of Node-RED itself by the way, one of the reasons I'm still using it after over a decade.

So, my advice, for what it is worth, is ignore all the noise you see on YouTube and elsewhere - those folk are often focussed on the latest "thing" and unpacking the darker, more complex parts of the web. But, for the most part, very few people ever need to get into that. And the basics are easy.

I liken it to learning to play the guitar. It is a really approachable instrument if you just pick it up and have a go. But, mastering it is a whole different ball-game. But most people who play guitar don't need to master it! The Beatles became famous knowing only a few chords!

To create a web page only requires a really basic, simplistic set of knowledge. UIBUILDER has already laid out that basic template for you. Take 10 minutes to get familiar with that template and you will soon be shocked at what you can do.

By the way, I count it as one of the best things I every did, as my career progressed into information enterprise architecture, was knowing how to knock up something in HTML/CSS/JavaScript. Amazing how you can impress people with a nice looking (but simple under the skin) web page. :rofl:

Coupled with Node-RED and some simple HTML and UIBUILDER, you can build custom data-driven web pages in minutes.

I made the move from D1 to D2 and yes it was painful but the result is far superior to the old dashboard. And of course there were several bugs, which to be fair were fixed by the developers once I raised issues for them..
Everything I could do with D1 I can now do with D2. And I'm using multiple instances in a production environment, so from my perspective D2 is production ready

I think the current state of Dashboard 2 is pretty impressive for this stage of development
Just to say I have no affiliation with flowfuse

And that's my tuppenceworth

Is it though? Has anyone made a statement on its future? Seems to me like things have come to a grinding halt.

Things have slowed for various reasons; but it isn't abandoned. It competes for time and resource against lots of other things. I'm very aware we need to get things moving again.

I too have found D2 to be buggy and not very user friendly. Seems we could have done this a bit better. But then again i"m on the outside looking in... For now I have resorted to sticking with D1 as it is mostly stable and the UI editor is better that D2's. Besides my app has grown so much over the last 6 years that in order to convert it would require vast amounts of time. And yes I've tried the converter. I may try again in the future but for now I'm sticking with D1. Not too worried about security as my installation is landlocked...

Cheers!