Looking for Beta Testers

I'm looking for a few beta testers to take a look at the most recent build of the plugin (you can download the latest build directly from GitHub.)

From the GitHub page, select "Clone or Download" and select "Download ZIP". This build has been running stably for me in development and on my production server. That said, there are a metric ton of changes, so proceed with caution and it's best to create backups of your current plugin settings and CSV files.

Here's a copy of the change log since the last stable release (v0.7.18). Please post any issues to this thread or to the GitHub Issues page.

Code: Select all

v0.7.27
- Reorders plugin device model list when editing device.
- Improves error logging.
- Cleans up XML attributes.

v0.7.26
- Revises chart device names for consistency.
- Fixes bug in battery chart device "KeyError: 'customSizeChart'. "

v0.7.25
- Adds choice to bar, line and scatter charts to suppress the plotting of
  individual data elements. The data and all settings are retained.

v0.7.24
- Synchronize self.pluginPrefs in closedPrefsConfigUi().

v0.7.23
- Adds choice to set bar width to make it simpler for first time device
  creation. If the value is set to zero, the plugin will attempt to set an
  attractive bar width automatically. Zero is now the new default for new bar
  chart devices.
- Changes menu items from "En/Disable All Devices" to "En/Disable All Plugin
  Devices".
- Changes default font for Calendar Devices to 12 pt (to better fit the default
  image size for non-retina screens.)
- Changes behavior of custom size chart checkbox so that now, custom size
  values will only be honored when the custom size checkbox is checked.
- Improves validation of plugin device configuration settings.
- Improves setting of Y1 limits and better handles condition where limits set
  improperly when all observations are the same value.
- Updates CSV device validation for duration (data to keep).
- Updates behavior so that if all CSV data are older than the time limit,
  the plugin will return the original data and warn the user.
- Refines grid settings for Polar Chart Devices
- Fixes bug in Line Chart devices where customizations (like best fit or fill)
  would result in an 'argument dimensions are incompatible' error.
- Fixes bug in default line color settings for Weather Forecast devices.
- Fixes bug in new CSV data source names that contain extended Unicode
  characters.
- Fixes bug in new and existing Polar Chart data source names that contain
  extended Unicode characters.
- Fixes bug where data header name was converted to 'NaN' while data quality
  repairs were made.
- Fixes bug where line chart best fit property incorrectly stored as a string.
  Will convert legacy devices in this state to boolean.
- Changes Python lists to tuples where possible to improve speed.
- Updates kDefaultPluginPrefs
- Code refinements.

v0.7.22
- Removes dependence on pandas library.
- Code refinements.

v0.7.21
- Removes default location from plugin preferences for data and chart save
  locations.

v0.7.20
- Improves resiliency in dealing with malformed CSV files.
- Adds new date formats for chart axes:
    Jan 16       [M D]
    Jan 16 2019  [M D Y]
    16 Jan       [D M]
    2019 Jan 16  [Y M D]
- [Line Devices] Hides the best fit color control until the "Plot Best Fit"
  checkbox is selected.
- Fixes bug in date/time format specifiers.
- Improves CSV item processing performance.
- Fixes bug in manual refresh Action.
- *Attempts* to work around pandas bug in rare csv write bug where the string
  '-01-01 00:00:00' is added to observation data.
- Refines plugin logging.
- Improves under-hood maintenance to ensure that the plugin and its devices
  are up to date for the installed version (plugin preference, device
  properties, etc.)
- Code refinements.

v0.7.19
- CSV Engine / Edit Data Item / Data Source control now returns a case-
  insensitive sorted list.
- Fixes bug in naming of CSV Engine data items that contain Unicode characters.
- Ups the default duration limit to no limit (where it should have been in the
  first place.)

I'm in

Backed Up

Loaded new Plugin

33 Graphs

3 Critical Errors that weren't there before

and the bug where it still performs a Fill even when fill unselected is still there - I wish this one would go away - every time it starts doing it the only option is to start again from scratch.

The Best Plot bug appears to have gone - i.e. carried son applying a Best Plot even after unselecting it....

Looking into the Critical Errors now.

OK First One

Any ideas - this is the simplest graph - a trend over the last three hours for one sensor

Code: Select all

Matplotlib Critical Error       [Chart Mini - Trend Outdoor Temp] RRuleLocator estimated to generate 116236 ticks from 1968-01-14 11:48:06.424057+00:00 to 2021-01-28 05:09:12.974423+00:00: exceeds Locator.MAXTICKS * 2 (2000) . See plugin log for more information.

Graph

Chart Mini - Trend Outdoors.png (1.28 KiB) Viewed 5334 times

Config

Chart 2 - Critical error - config.jpeg (164.67 KiB) Viewed 5334 times

Trend - Temp Outdoors.csv

CSV File

(417 Bytes) Downloaded 195 times

Thanks very much. I have upgrade debug logging considerably--so much so that some errors may be reported as critical where before they were passed. The debugging should be descriptive in the Indigo Events log and specific (line number, etc.) in the plugin private debug log.

You'll have to be more specific on the Fill bug -- I worked that one over pretty extensively in this update so it may be that you need to open the device config and then save (without making changes) to ensure the changes are promulgated. I had hoped that they would do that without interaction.

Thanks for the help!
Dave

OK First One

Any ideas - this is the simplest graph - a trend over the last three hours for one sensor

Have a look at the underlying CSV data -- that error is being reported by Matplotlib and it's saying that there's way too many ticks to plot the chart. It's saying that the observations are going back to 1968 -- which is extremely unlikely.

The csv is attached to report only 12 readings

Number 2
Fill still occurs after un selecting it
Have Start stopped plugin
Have opened - changed nothing and saved.

Still doing it


Fill Bug

Chart - Key Temperature Readings.png (44.82 KiB) Viewed 5327 times

Graph - Fill Bug - Config 1.jpeg (162.61 KiB) Viewed 5327 times

Graph - Fill Bug - Config 2.jpeg (150.26 KiB) Viewed 5327 times

Graph - Fill Bug - Config 3.jpeg (162.53 KiB) Viewed 5327 times

HA

I have the fill showing in White so it shows up nicely on my Control page

I've changed it to something that shows

UPDATED - added config


Number 3

Code: Select all

Matplotlib Critical Error       [Chart - Power Tool Charger] to_rgba: Invalid rgba arg "#line1Co"
to_rgb: Invalid rgb arg "#line1Co"
invalid hex color string "#line1co". See plugin log for more information.

Config

Chart - Bug 3 - Invalid rgba.jpeg (149.25 KiB) Viewed 5313 times

It seems to dislike two charts I have that currently only 0.0 for data

Both attached below
They are useful so I can track that charging devices are fully charged and that the power was turned off to the mains for them.

Currently neither have been charged in the last three days so 0.0 for all entries - this error did not occur in last plugin.

*** UPDATED *** - Deleted Source CSV file and error stopped once MatPlotLib recreated the file and populated with a reading


Bug Number Three rgba error

A few things happened at the same time that altered the outcome

I am now getting this error

Code: Select all

Matplotlib Critical Error       [Chart - Power Tool Charger] list index out of range. See plugin log for more information.


The things that changed were

1. Edited Chart - changed colour by selecting from The Crayons 'swatches' 3rd colour selector along from left and chose a colour for the actual pictures of crayons - not sure if this is meaningful but I saw reference to rgba thought selecting from a different palette may affect things.?

2. I had turned on the tool Charger and it had 1 element of data in it at the end - When drawing the graph no errors at all.

3. Deleted last line of csv with the single actual data - the preceding hundreds of 0.0 only remained. On redrawing graph got he above error.

Bug Numer 1 - Chart Mini - Trend Outdoor Temp

I can't see any data before 3 hours ago let alone 1968

It is a file created by MatPlotLib

Code: Select all

Timestamp,Trend - Temp Outdoors
2019-02-10 15:11:47.179000,5.4
2019-02-10 15:26:49.486000,5.5
2019-02-10 15:41:50.470000,5.4
2019-02-10 15:57:02.155000,5.4
2019-02-10 16:12:04.260000,5.4
2019-02-10 16:27:12.018000,5.4
2019-02-10 16:42:17.200569,5.2
2019-02-10 16:57:19.398478,5.1
2019-02-10 17:12:32.554454,5.0
2019-02-10 17:27:38.429255,4.9
2019-02-10 17:42:41.082953,4.9
2019-02-10 17:57:41.524539,4.8


I have renamed the CSV file now and am, rightly getting...

Code: Select all

   Matplotlib Critical Error       [Chart Mini - Trend Outdoor Temp] chart updated with errors.

My guess is that when the next read - every 15 mins it reset itself and should be ok. _ will let you know.

OK that is my input up to date

I use this Graphing tool extensively to display trends and allow me to investigate troublesome devices, like the Spirit Radiator TRVs - I was able using MatPlotLib to graph everything and work out how they were operating

So many thanks for all your hard work on this - VERY MUCH APPRECIATED!

OK Update on Bug Numer 1 - Chart Mini - Trend Outdoor Temp

While csv missing getting

Matplotlib Critical Error [Chart Mini - Trend Outdoor Temp] chart updated with errors.

Code: Select all

MatPlotLib has recreated csv file with one entry

Code: Select all

Timestamp,Trend - Temp Outdoors
2019-02-10 18:13:01.423924,4.7

File attached also so you can re-create your end

and I get this error

Code: Select all

Matplotlib Critical Error       [Chart Mini - Trend Outdoor Temp] RRuleLocator estimated to generate 116236 ticks from 1968-01-14 11:45:04.743047+00:00 to 2021-01-28 06:27:56.680877+00:00: exceeds Locator.MAXTICKS * 2 (2000) . See plugin log for more information.

I am at loss now

No other ideas to try

Thanks very much for the input, but I'm afraid I'm having a bit of a tough time tracking your reports. It would be helpful to see the errors that are posted to the plugin log file when each error appears. Those entries show when the errors hit and what's going on with the error. When multiple things change at the same time, it's very hard for me to recreate what you're seeing without that log information.

It sounds like there are three things going on.

1. Fill when fill turned off.
2. Too many ticks error - this may be related to all data being zero.
3. rgba error with color "Invalid rgb arg".

I'll see what I can do to recreate the errors you're seeing with the data you've provided. If you can share the plugin private log file it would be super helpful.

Code: Select all

/Library/Application Support/Perceptive Automation/Indigo 7.2/Logs/com.fogbert.indigoplugin.matplotlib/plugin.log

I really appreciate the help!