Incorrect Custom Float Property Value Readout #49304

New Issue

Darren Waschow · 2016-09-08T22:53:22+02:00

Darren Waschow commented

2016-09-08 22:53:22 +02:00

System Information
Windows 7 Enterprise 64 w/ NVIDIA Quadro K2100M

Blender Version
Broken: 2.77a abf6f08 and 2.78 RC1 0e5089c
Worked: (optional)

Short description of error
Assigning custom float property values to objects stores incorrect values. I've tried by creating properties via the UI as well as with python code:

import bpy
ob = bpy.context.active_object
ob['property'] = 987654.321

Doing so (in both cases) the property is set to the correct value internally, but the UI reports the value as 987654.313.

Other numbers are also coming up with incorrect decimal parts of the value.

Exact steps for others to reproduce the error
Create a cube.
Add a new Custom Property using the UI or with the above code run via a text window.
Check property value via Python console (it'll be correct) and then look in the UI and see the incorrect value.

**System Information** Windows 7 Enterprise 64 w/ NVIDIA Quadro K2100M **Blender Version** Broken: 2.77a abf6f08 and 2.78 RC1 0e5089c Worked: (optional) **Short description of error** Assigning custom float property values to objects stores incorrect values. I've tried by creating properties via the UI as well as with python code: import bpy ob = bpy.context.active_object ob['property'] = 987654.321 Doing so (in both cases) the property is set to the correct value internally, but the UI reports the value as 987654.**313**. Other numbers are also coming up with incorrect decimal parts of the value. **Exact steps for others to reproduce the error** Create a cube. Add a new Custom Property using the UI or with the above code run via a text window. Check property value via Python console (it'll be correct) and then look in the UI and see the incorrect value.

Darren Waschow commented

2016-09-08 22:53:22 +02:00

Changed status to: 'Open'

Darren Waschow commented

2016-09-08 22:53:22 +02:00

Added subscriber: @DarrenWaschow

Brendon Murphy commented

2016-09-09 02:53:04 +02:00

Added subscriber: @BrendonMurphy

Brendon Murphy commented

2016-09-09 02:53:04 +02:00

hi, tested this out & found the error occurs after more than 8 digits are added, less than 8 digits & there's no difference, more than 8 & there is wrong comparison.

Bastien Montagne commented

2016-09-09 14:13:35 +02:00

Added subscriber: @mont29

Bastien Montagne commented

2016-09-09 14:13:35 +02:00

Changed status from 'Open' to: 'Archived'

Bastien Montagne closed this issue

2016-09-09 14:13:35 +02:00

Bastien Montagne self-assigned this 2016-09-09 14:13:35 +02:00

Bastien Montagne commented

2016-09-09 14:13:35 +02:00

Thanks for the report, but no bug here, that’s merely well-known floating point precision issue, you only have about 7 digits usable on 32bit floats, nothing to be done here.

Darren Waschow commented

2016-09-09 14:55:19 +02:00

Bastien:

I'm not sure your explanation addresses what's actually happening here. The value stored in the object's property is correct. It's just the value that's shown in the UI under Custom Properties that's incorrect.

If you assign a floating point value of, say, 987654.321 to an object property, you can then query that value using ob['property'] in the Python console and it's correct. But the UI shows the value as 987654.313.

See the attached screen shot.

Thanks!

Bastien: I'm not sure your explanation addresses what's actually happening here. The value stored in the object's property is correct. It's just the value that's shown in the UI under Custom Properties that's incorrect. If you assign a floating point value of, say, **987654.321** to an object property, you can then query that value using ob['property'] in the Python console and it's correct. But the UI shows the value as **987654.313**. See the attached screen shot. Thanks! ![Property2.png](https://archive.blender.org/developer/F356609/Property2.png)

Bastien Montagne commented

2016-09-09 16:00:27 +02:00

@DarrenWaschow That’s precisely the point - our UI code does not draw raw float values, for various reasons, so it ends up applying some transformations to it (see UI_calc_float_precision function in code), hence the float precision issue…

@DarrenWaschow That’s precisely the point - our UI code does not draw raw float values, for various reasons, so it ends up applying some transformations to it (see `UI_calc_float_precision` function in code), hence the float precision issue…

Sign in to join this conversation.

No Label

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

Incorrect Custom Float Property Value Readout #49304