-
Notifications
You must be signed in to change notification settings - Fork 2
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Test report for RC3 #19
Comments
So I had not been active with testing as my unit was completely non-functional on wifi. Seemed to be a problem with 4.1.5, but I finally got a chance to be stubborn and power cycle it a few dozen times, so I could get 4.1.7 loaded onto it. We'll see if it behaves from here on out, wish me luck. Testing things now...
|
@fhteagle, there's now daily builds with new ui here: https://github.com/OpenEVSE/ESP32_WiFi_V4.x/releases/tag/v2_gui?fbclid=IwAR1Ty7Wr735bHMQkGC1wAZp0_K4FcNVNFThaVxQ9_hIbti59UYKUKpWAVb8 Do not put 4.1.7 it has a bug with MQTT making wifi drops and more. |
@fhteagle , I have fixed the time setting issue, if you can test that now everything is workoing ok for you |
At release: Agreed that your nightly is more stable than the official 4.1.7 . Wifi is working way better than it has since about 4.1.3 or so. Any chance we can get the version incremented in the UI so we can see we are not on "stock" 4.1.7 anymore? A version string like 4.1.7-rc3-d98eb7d4 would help to assure the user that their uploaded nightly build firmware did in fact get applied. Visual glitch: on Vivaldi on Android, the top part of the status bar is cutoff. The two icons at the top left for charge status and vehicle connection status, the OpenEVSE logo, and the time block at the upper right all cannot be seen. Firefox Focus, stock Chrome browser on Android display as expected. On Opera on Android, the "Connection Error" banner shows all the time, even when tabs load with correct data. On the history tab, I cannot see a way to read the content of an Error event line. For example, I had a false "No Ground" detection a few days ago. I see the red caution triangle with ! icon, but no way to see that that history tab line is a No Ground event. I did test time setting. Manual seems to work reliably, but the time jumps to UTC when the Manual option is first selected. Local time option works reliably for me. Setting UTC time zone when NTP is selected does technically work, but feels a bit of a hack to me. It works, just could be a bit more slick. If time zone is not going to be retained, maybe ask the user the correct timezone immediately, or focus and highlight the timezone entry box? I will keep exploring the UI and see if I can find anymore issues. Thanks. |
Also you'll notice there's UX difference between some forms. I have started to rework them, like HTTP ,MQTT m, Time and RFID. They will all be done the same way without save button ( but only if really necessary ) edit: I have just commited the error code display on Logs and Status bar, thanks for remind me it |
Updated to the nightly build from 5 hours ago. Tooltips on icons in history works, merci beaucoup. However, I just noticed that the date/time column of history appears truncated: Using OEM Android 12 on OnePlus 7T: Still seeing the truncated status block in Vivaldi 5.6. Same with the connection error banner on Opera 72.5.3767.69342. I can start checking other phones if we need to track these display bugs down further. On MQTT settings, the "Show" tickbox does not reveal the password. |
Also, on the history tab, rightmost column, can we get the Temp to be a consistent precision? I really do not care if it is an integer only (17) or one decimal place (17.2), but having them all the same precision makes the column look cleaner. |
Thanks @fhteagle , I think I've covered all your issue but the Opera websocket disconnection error I can't reproduce. Edit: I've found it, it's the VPN prolly activated in your Opera, seems to cause trouble with websockets . I have posted a bug report on Opera team. Btw, why using Opera in 2023? |
I tested again on an S9 with Android 10. Could not reproduce either the Vivaldi or Opera bugs on that phone. So both UI display glitches may be a one off issue with my OnePlus 7T. I have a friend who sends me all kinds of current events and political stuff, some of it useful alternative news, but some of it is straight up crazy conspiracy theory stuff. I use Opera Privacy Tabs to load the links from him, so it does not get associated with my usual digital footprint lol. Just thought I would try the GUIv2 out there as well (not in a privacy tab of course), just for testing purposes. Agreed that is not a high market share browser and probably not worth spending too much time worrying about. |
Hey I have made a dirty hack to handle the OpenEVSE time bug can you try latest build? I'll remove that when @jeremypoulter solve it on api I have also removed the set to browser time, it nows automatically set to local time when you select manual, |
Dirty hack seems to operate as expected from every browser I checked. Agreed that the first status block update with the completely random time is odd, possibly confusing for the user. Is there a way to mask the first wrong time refresh and keep it from displaying? Perhaps replace with "Updating time...."? |
I could make the websocket ignore time property for the first receive time update. But that's a really dirty hack 😁 |
After GUI-triggered restart of EVSE and Wifi board, time gets stuck at UTC until I do a manual change of time and back to NTP. Probably a result of the "dirty hack". Not sure it is worth fixing, though, maybe just wait for the API to get fixed properly.... |
Other really really really minor thing, I am testing Homeassistant controlling the OpenEVSE through an integration that uses primarily HTTP API. When the "max amps" is selected in HomeAssistant, sometimes the GUI does not update the max amps on the main charging tab, even after a few of the 30 second update ticks have gone by. If it does update, I get an extra zero in the front of single digit amp values. After refreshing the browser shows the updated value from HomeAssistant automation 100% of the time though. So I think it is just the GUI not getting an update somehow. Note to self: do more digging to see if the integration is updating the wrong soft vs hard max amps variable.......... |
I wonder why the max_current value is modifiable through HA. This is a one time setup. Shouldn't they adjust charge_current instead? |
Home assistant should probably expose both as you could use it to implement many use cases, however the charge_current should be the main one that shows up by default, the max current should probably be hidden by default. |
If HA use max_current_soft from /config, it then should display in the UI slider max pos as it should download automatically the new /config if there's a change. |
This is what I was trying to confirm about in this issue over there: |
Retested GUIv2 release from 3 hours ago, the following bug remains: On MQTT settings, the "Show" tickbox does not reveal the password. |
it's normal, it only reveal while you have typed it yourself. The api doesn't send the password so it stays hidden |
Okay, suggestions to handle that behavior more elegantly: a. hide the Show tickbox if there is nothing to show yet, and/or I think both of these would more clearly communicate that the GUI does not know what the password is, so do not expect to be able to show it... Just grabbed the latest nightly build, planning on going through everything I can test on my setup very thoroughly today. |
Thanks for the input. I consider hiding the tickbox yes, it's on my list. |
For the v1 UI we show nothing when the show password is selected on an existing password. We do this because it is likely that the user wants to enter a new password |
I have considered that,but then we don't know a password has been set before. |
I am not seeing any other showstopper bugs in GUI v2 after a couple of weeks of pretty thorough testing on the nightlies, trying to break things, etc. Great work @KipK on this. |
You can download the latest build, there's many changes and it needs to be tested ;) |
I've been pulling each new build as it shows up here: https://github.com/OpenEVSE/ESP32_WiFi_V4.x/releases/tag/v2_gui Or is there a different branch that needs more testing? |
No it is this one. Have you tested the new Limit engine? |
Just noticed the new limit box on the main tab. I will try it out and report back. |
Hold on, I've just tested the generated fw on githb and it's not the good version, UI is ok but not the latest fw. I tell you when it's ok |
my PR was merged yesterday ( OpenEVSE/openevse_esp32_firmware#535 ), I've pushed the Limit branch on UI too, but the UI_V2 branch of wifi fw has not been updated yet. So it generated a build with the Limit UI but without the feature on fw :) |
Will the limit engine controlled by Wifi firmware features require flashing the OpenEVSE firmware itself, or can I test against 7.1.3? |
Nope It's a wifi fw feature only |
good to go now . Thanks for your help |
Found a quirk: If session is already in progress, and a time or energy limit is set below values already accumulated, the session goes straight to disable. This could be problematic for users and especially automations that are not aware of existing session amounts. Potential solutions:
I am not sure which option(s) is/are most beneficial to users. I can make reasonable arguments in my head for each of them. How did GUI v1 and the old limit system handle this situation? Also, rounding needed on time remaining figure: |
Also, is it not possible to set both a time and energy limit simultaneously? Is this by design? |
hehe I was just correcting this display bug. I've pushed the fix, build in progress. |
yes for now only one limit. I thought about allowing multiple limits, but I've not found real use case but gadget ones. And it complicates the UX. |
Robot Icon means there's an active service controlling the states of the EVSE. Limits is only made to disable the state. It's not part of automation features as it's still enabled when state is forced to active with Manual override. But seeing the Enable/Disable buttons not selected , I get your problem. I have to think about that, but you're right, I should enable the robot Icon then. It just bugs me because it goes against what I've planned for it. Why not just set the button to disabled state and no robot ? Would this be confusing in UX perspective ? |
Can you post this first & last issues on github ? I'm leaving for 10 days and don't want to forget. |
@KipK - Time limit display bug fix verified, thanks, looks much better. "Why not just set the button to disabled state and no robot ?" That makes sense to me too. If limit has been triggered, OpenEVSE is acting as disabled, so highlighting the disabled button with red background communicates that well. I'm not sure which of my observed behavior comments you were wanting to have split to separate issue items. Happy to write-up the split, just need to know which ones. |
I mean the 2 bugs / suggestions reports you've made but you can also create one for the simultaneous limits. It's easier to track them separated. |
I have pushed RC3 here :
https://github.com/KipK/openevse-gui-v2/releases/tag/1.0.0-rc3
There's a binary build for openevse_wifi_v1 module that can be installed safely.
@fhteagle, @jeremypoulter If you can give a look.
The text was updated successfully, but these errors were encountered: