Predictive completion not working with completion scripts generated by the cobra library

[shellwhale]

Prerequisites

• [X] Write a descriptive title.
• [X] Make sure you are able to repro it on the latest version
• [X] Search the existing issues.

Steps to reproduce

1. Import the CompletionPredictor module with Import-Module -Name CompletionPredictor
2. Set the PredictionSource option to HistoryAndPlugin with Set-PSReadLineOption -PredictionSource HistoryAndPlugin
3. Generate and execute a cobra-generated completion script (e.g., kubectl completion powershell | Out-String | Invoke-Expression)
4. Type a command that uses the completion script (e.g., kubectl get pods) and try to use predictive completion.

Expected behavior

~ ❯ kubectl get
> kubectl get pods                                                                         [Completion]
> kubectl get csr                                                                          [Completion]
> kubectl get nodes                                                                        [Completion]
> ...

Actual behavior

~ ❯ kubectl get
> kubectl get pods                                                                         [History]

Error details

No response

Environment data

Name                           Value
----                           -----
PSVersion                      7.3.4
PSEdition                      Core
GitCommitId                    7.3.4
OS                             Microsoft Windows 10.0.19045
Platform                       Win32NT
PSCompatibleVersions           {1.0, 2.0, 3.0, 4.0…}
PSRemotingProtocolVersion      2.3
SerializationVersion           1.1.0.1
WSManStackVersion              3.0

Version

0.1.0

Visuals

No response

[Daydreamer-riri]

‌‌‌‌‌‌Hi @daxian-dbw, currently, all custom completions and executable files do not prompt. Is there a way to achieve this functionality?

[daxian-dbw]

Unfortunately, the completion for native commands doesn't work with completion predictor today because they are too slow. For a predictor to work with PSReadLine today, it needs to return results within 20ms, which is impossibly for most of native command completers. We have a work item in PSReadLine to allow longer timeout by separating the prediction rendering from the user typing rendering, but we haven't got to it yet.

[Daydreamer-riri]

Thanks for your reply! I've just seen the latest vscode terminal integration do something similar, but I'm using Windows Terminal more often than not.