chore(release): Bump to version 0.5.3

[niheaven]

Changelog

Bug Fixes

• scoop-download|install|update: Fallback to default downloader when aria2 fails (#4292)
• decompress: Expand-7zipArchive only delete temp dir / $extractDir if it is empty (#6092)
• decompress: Replace deprecated 7ZIPEXTRACT_USE_EXTERNAL config with USE_EXTERNAL_7ZIP (#6327)
• commands: Handling broken aliases (#6141)
• shim: Do not suppress stderr, properly check wslpath/cygpath command first (#6114)
• scoop-bucket: Add missing import for no_junction envs (#6181)

Code Refactoring

• download: Move download-related functions to 'download.ps1' (#6095)

Performance Improvements

• shim: Update kiennq-shim to v3.1.2 (#6261)

Close Issues

• Closes #6011
• Closes #4472
• Closes #6112
• Closes #6180
• Closes #6225

Credits

Thanks to the following contributors (ordered by PR merging time):

@o-l-a-v @niheaven @kiennq @HUMORCE @sitiom @chawyehsu @basselworkforce

[niheaven]

@ScoopInstaller/maintainers Anyone could review this PR?

[tech189]

Hi @ScoopInstaller/maintainers, let's have a poll before we start allowing the use of AI coding tools in the Scoop project.

Personally I'd be against using them. Bigger projects like curl are already struggling with AI generated bug reports - see this Ars Technica article - and they can't be trusted to write safe code. There have also been a lot of people wanting the option to turn them off per repository.

👍 if we should use them, 👎 if not!

[Lutra-Fs]

Hi @ScoopInstaller/maintainers, let's have a poll before we start allowing the use of AI coding tools in the Scoop project.

Personally I'd be against using them. Bigger projects like curl are already struggling with AI generated bug reports - see this Ars Technica article - and they can't be trusted to write safe code. There have also been a lot of people wanting the option to turn them off per repository.

👍 if we should use them, 👎 if not!

It’s definitely an important conversation to have while AI is becoming more and more powerful and dangerous.

Personally, I believe AI-assisted tools (especially code review tools) are not inherently bad and can offer useful insights when used responsibly.

As far as I know, GitHub Copilot Agent (which can act more autonomously) is disabled by default, and although GitHub allows organization level settings, there’s currently no reliable way to enforce those settings on contributors’ machines. Even if Copilot is disabled, users can still access AI support through local tools like Cursor or Cline.

I do agree that AI often lacks full context across files and can suggest speculative issues without proper validation — that’s definitely a concern, especially when it leads to bug reports with no reproduction steps. Those shouldn’t be encouraged.

That said, with good engineering discipline, AI tools have also helped researchers uncover real issues — including a recent example involving Linux kernel bugs: https://news.ycombinator.com/item?id=44081338

In my view, banning AI tools entirely might not be practical or necessary. Instead, we could focus on setting expectations around quality: • no AI-generated PRs without human collaboration (coding,testing) and verification • no unverifiable bug reports • and encourage responsible, transparent usage.

[Lutra-Fs]

The AI issue/PR flood is something GitHub has to fight against. I mean, those malicious submits do add a lot of workloads to maintainers, whether or not we do anything/nothing to them; their existence themselves is a workload.

[WordlessEcho]

Also see hotfix #6435 #6436