john-guo / simply-view-image-for-python-opencv-debugging Goto Github PK

Hey, I have been searching for something like this extension, great work.
There is a bug when the image is in range 0..1 floats, it is not saved correctly.
There should probably be some checking: if img.dtype = np.float32 or np.float64 then img * 255.0

I am not seeing the yellow bulb anymore

Hello, thanks for your great extension that help me a lot. However, the extension seems doesn't work correctly.

I have tried 2 image-debugging extensions, this one and another python-image-preview, and strange things happen:

When I only enable this extension, there is no lightbulb when I debug.

While I enable both this extension and the python-image-preview, the two extensions work well.

So where is the problem?

Hi John! Thanks for this great extension!

I have one problem though, it doesn't seem to we working when I'm connecting via SSH to a server and use X11 Forwarding. I can see the images when I run cv2.imshow() but I don't get the the preview option when hovering on the image variable.

thanks for this extension this is amazing.
Would it be also possible to make it to be able to display the image when I do it in debug console.
I mean, lets say i have:

import torch as np
img = np.random.rand(100,100,3)

now I just can display the image (img) in the main file. would it be possible that I copy and paste np.random.rand(100,100,3) in the debug console like simple_view(np.random.rand(100,100,3)) and it display the image. simple_view is gonna be a function from your code that you tell us to run to see the image.

Thanks for this extension anyways :)

Hello,

The extension was working amazing a couple weeks ago and now it has stopped working.
When I select the image variable, it doesn't show up the light bulb for image view.

VScode version: 1.63 (November)

Hi,

I am debugging the remote python code through wsl, and the plugin can not show the image during debugging the python code.

The logs says:

[2021-01-22 11:53:54.619] [exthost] [error] [johnguo.simply-view-image-for-python-opencv-debugging] provider FAILED
[2021-01-22 11:53:54.627] [exthost] [error] Error: Unable to find thread to evaluate variable reference.
at t.RawDebugSession.handleErrorResponse (file:///C:/Users/SHData/AppData/Local/Programs/Microsoft VS Code/resources/app/out/vs/workbench/workbench.desktop.main.js:3369:750)
at file:///C:/Users/SHData/AppData/Local/Programs/Microsoft VS Code/resources/app/out/vs/workbench/workbench.desktop.main.js:3369:55

Pls review.

Thank you !

Hello,

Thank you so much for the great plugin. I think if you advertise it correctly, it will be helping many developers!

I wonder if you could add support for TF2 tensor arrays too. Currently, I have to manually convert tf tensors to numpy tensors with .numpy() to be able to visualize them.

Many thanks

Hey there. For the last few days I have been experimenting with something called WebView. It essentially allows HTML rendering inside vscode.
There have been many requests over the years for vscode community to allow inline WebView render into editor.
Here is something similar to what we want: microsoft/vscode#41775
VScode has added and merged a code inset feature here: microsoft/vscode#66418
And here is their progress for WebviewEditorInset API: microsoft/vscode#85682

There is however a drawback at this moment:

1. WebviewEditorInset is a proposed API, meaning that I don't know if this can be used is its not officially into vscode. It is like an experimental feature. But i'm pretty sure they will implement this after half a year or year (yeah, that a lot), but this has been a really requested feature and they also need it for julia lanuguage extension and jupyter notebooks.
2. Current solution requires to copy a vscode.proposed.d.ts file for this to work
3. VS Code insiders edition is also required for this

Heres a demo that I have managed to get working:

There are a lot of work to be done here, but I think it looks cool.

Also, you mentioned that you have to create a /tmp directory for this, check out what storagePath is in vscode.ExtensionContext in https://code.visualstudio.com/api/references/vscode-api#ExtensionContext
I am using this directory instead of /tmp, it looks like this: /home/maris/.config/Code - Insiders/User/workspaceStorage/RANDOM-HASH/johnguo.simply-view-image-for-python-opencv-debugging/

Is this the direction we can shift instead of opening images into a new pane? We can also try to combine them, or make this as a separate experimental branch.