refactor(scoop-list): Allow list manipulation

[rashil2000]

Description

Return PSObject instead of printing to terminal.

Motivation and Context

Allows manipulation, sorting etc. of the list with the help of PowerShell's pipelines.

Closes #3162
Closes #4138
Closes #4179

How Has This Been Tested?

Before

After

More features (not possible before)

Homebrew-style output:

Filter apps from a bucket:

Checklist:

• I have read the Contributing Guide.
• I have updated the documentation accordingly.
• I have updated the tests accordingly.

[rashil2000]

In the next PR I will add lastwritetime field so that we can sort the list by last updated/installed time

[niheaven]

Source or Bucket? I think Bucket is better for Scoop.

[rashil2000]

Source because it can be a local file path or a remote URL also, so Bucket won't fit there

[rashil2000]

I just realized that this PR (and the previous one for buckets by @HUMORCE) does not work in Cmd and Bash. No output is shown, unless we remove the exit $lastexitcode part from these lines:

Scoop/lib/core.ps1

Line 645 in 7967905

" pwsh -noprofile -ex unrestricted -command `"& '$resolved_path' $arg %args%;exit `$lastexitcode`"",

Scoop/lib/core.ps1

Line 656 in 7967905

" pwsh.exe -noprofile -ex unrestricted -command `"& '$resolved_path' $arg $@;exit \`$lastexitcode`"",

@niheaven any idea why these are essential, and whether we can remove them?

[niheaven]

This line has been added here for a long time and I must check original codes.

[niheaven]

Found an old explanation of this exit $LastExitCode (#339 (comment)), Maybe it's better to find out why there is no output...

[niheaven]

Some suggestions: For this generated manifest, tell user this is Generated and a manifest location is more better. This needs further parsing for manifest location with workspace.

[rashil2000]

Found an old explanation of this exit $LastExitCode (#339 (comment)), Maybe it's better to find out why there is no output...

I kinda agree with nightroman there. I didn't fully understand the explanation luke gave (also there are no examples in his comment to reproduce the problem if we remove it).

Maybe we just put exit %errorlevel% at the end of the .cmd file?

[rashil2000]

I just checked. If you run the command powershell -nop -c "exit 1" in Cmd, it will show the correct exitcode. The exit $LastExitCode in the line is redundant IMO.

[niheaven]

See this (PS7) and this (PS5)

The process exit code is determined by status of the last (executed) command within the script block. The exit code is 0 when $? is $true or 1 when $? is $false. If the last command is an external program or a PowerShell script that explicitly sets an exit code other than 0 or 1, that exit code is converted to 1 for process exit code. To preserve the specific exit code, add exit $LASTEXITCODE to your command string or script block.

And in @nightroman's repo, we may use $Global:LASTEXITCODE in shim: https://github.com/nightroman/PowerShellTraps/tree/master/Basic/LastExitCode

[rashil2000]

Instead of using -Command parameter, we can use -File, which does not suffer from the exitcode limitation described in the Microsoft Docs. Any exit code is returned to the Cmd shell as is.

powershell -noprofile -ex unrestricted -file "D:\Data\Scoop\apps\scoop\current\bin\scoop.ps1"  %args%

(I'm not even sure why we use -Command in the first place, when -File parameter exists exactly for this purpose).

[niheaven]

Instead of using -Command parameter, we can use -File, which does not suffer from the exitcode limitation described in the Microsoft Docs. Any exit code is returned to the Cmd shell as is.

powershell -noprofile -ex unrestricted -file "D:\Data\Scoop\apps\scoop\current\bin\scoop.ps1"  %args%

(I'm not even sure why we use -Command in the first place, when -File parameter exists exactly for this purpose).

PS5 doesn't document exit code for -File clearly, and this code block is written about 8 years ago while PS's version is 2. So I'm not sure if it will output correct exit code in current PowerShell edition (5.1 and 7).

If -File works, it is better, IMO.

[rashil2000]

I tested in PS 5.1 before.

Just now, I tested in PS 7.1.2.

Both work nicely 👍

[niheaven]

Not only scoop itself, but also other apps with ps1 shim? If so, let's use -File and test in develop branch.

Emm, it's better to open a new PR for such changes.

[HUMORCE]

I just realized that this PR (and the previous one for buckets by @HUMORCE) does not work in Cmd and Bash. No output is shown, unless we remove the exit $lastexitcode part from these lines

switch parameter for return? nothing else comes to mind at this time.

[rashil2000]

I just realized that this PR (and the previous one for buckets by @HUMORCE) does not work in Cmd and Bash. No output is shown, unless we remove the exit $lastexitcode part from these lines

switch parameter for return? nothing else comes to mind at this time.

It will work once #4721 is merged

[HUMORCE]

It will work once #4721 is merged

i seem to have missed some comments in this thread on gh mobile.

[niheaven]

Some suggestions: For this generated manifest, tell user this is Generated and a manifest location is more better. This needs further parsing for manifest location with workspace.

What about this? Or maybe another PR?

[rashil2000]

I have not added any new features in this PR. Scoop list shows the same information as it did before.

Each new feature should be a new PR IMO