Cannot scroll up after displaying license and error message if I declined

[jkrech]

Windows cmd:
cpackget add Arm-Packs::Unity
Pressing key 'd'

E: user does not agree with the pack's license
E: user does not agree with the pack's license

C:\Open-CMSIS-Pack\ctools\bin>

I cannot scroll further up. The first line is the E: which also gets duplicated
Note I am not sure I would consider this an error, as the operation selected by the user was executed successfully.

[chaws]

I tested on Linux and scrolling seems fine after declining the license. Let me report back after replicating this issue on Window.

[lud0v1c]

Tried this out with Powershell and it happens as described, after declining the error message gets duplicated and the current screen gets wiped.. Investigating why

[chaws]

I have done some experimenting and we the only workaround I could work out is opening a new cmd window just with the license. Then after user's response, the window would close out and cpackget would continue from there.

Does that seem reasonable to you @jkrech ?

[jkrech]

@chaws I am not sure how this could work if I am in a terminal session to the machine? Where would that other window pop up? I think you would need to run the cmd in the same window?

[chaws]

@jkrech this would only be applied if you're running a GUI. The idea is:

1. Window 1 (git-bash or cmd): cpackget add <pack-with-license>.pack
2. Window 2 will be a new cmd terminal, displaying only the license window like the image below. Meanwhile, cpackget on Window 1 will be waiting for user input on Window 2
3. User accepts/declines/extracts the license
4. Window 2 close out
5. Window 1 continue with installation/extraction. Afterwards, scrolling will not be affected.

[chaws]

I talked to Luis, we both did some research on why scrolling gets messy, but we couldn't find the reason why unfortunately. If I was to guess, it is due to some hidden configuration on the terminal. Note that this does not happen in git-bash.

[jkrech]

Assuming we can rely on the detection of "running with GUI", this makes sense then. But it does not change the problem for the none-GUI mode then, right?

[chaws]

The interactive-mode detection on cpackget is what it seems the common practice. While it has a false positive on git-bash (hence #95), it shouldn't try to pop up a new window on CI, for instance.

This behavior will only be applied on Windows platforms.

[lud0v1c]

Error message is now a normal info message (eaa7f7e).