livecheck: add --extract-plist

[razvanazamfirei]

• Have you followed the guidelines in our Contributing document?
• Have you checked to ensure there aren't other open Pull Requests for the same change?
• Have you added an explanation of what your changes do and why you'd like us to include them?
• Have you written new tests for your changes? Here's an example.
• Have you successfully run brew style with your changes locally?
• Have you successfully run brew typecheck with your changes locally?
• Have you successfully run brew tests with your changes locally?

---

Running brew livecheck --cask --eval-all downloads all casks using the ExtractPlist strategy. This takes up a significant amount (~20GB) of bandwidth and cache space.

If this flag is added, it will allow maintainers to check taps more quickly and with less resources consumed.

[apainintheneck]

Out of curiosity I looked it up and this would affect 275 core casks. Outside of memory savings is this faster as well?

Either way this seems reasonable to me.

[bevanjkay]

Out of curiosity I looked it up and this would affect 275 core casks. Outside of memory savings is this faster as well?

@apainintheneck
It is significantly faster to run brew livecheck if you skip the extract_plist.
The strategy downloads the binary and checks the contents for a version number, whereas most of the time we are checking HTML webpage contents, or a curl response header.
As some of the files are quite large the network requests can take some time to complete. Note that we only use this strategy when all other alternatives are not possible.

I have my own script locally that runs livecheck on a tap after filtering out the extract_plist casks, which could be replaced with this flag.

[MikeMcQuaid]

Running brew livecheck --cask --eval-all downloads all casks using the ExtractPlist strategy. This takes up a significant amount (~20GB) of bandwidth and cache space.

I really appreciate the PR (great work @razvanazamfirei!) but: the above described behaviour makes no sense.

I'd argue instead that it should instead be one of:

• --extract-plist is added instead so it's opt-in
• --eval-all never attempts to extract plists
• something else where the above behaviour doesn't manifest

[reitermarkus]

Maybe this could work similarly to audit with --only and --except flags, so it can also be used more generically to validate changes e.g. when changing a strategy.

[apainintheneck]

• --extract-plist is added instead so it's opt-in

I like this idea. Do we have to worry about backwards compatibility with this command though. If so, we'll have to go through the whole deprecation cycle and the way it's currently written would be the fastest way to get access to it. It seems like more of an internal command so we're probably fine though.

[MikeMcQuaid]

Do we have to worry about backwards compatibility with this command though.

100% not.

It seems like more of an internal command so we're probably fine though.

Agreed.

[MikeMcQuaid]

We could even perhaps have a documented policy that some/all developer commands do not require deprecation periods at all.

[razvanazamfirei]

Thank you all for the suggestions! I've reworked this.

1. The default behavior has been changed to skip ExtractPlist casks.
2. Single cask checks maintain current behavior.
3. Added --extract-plist flag to opt-in to the livecheck strategy.

[apainintheneck]

The code looks good to me now. I left a nit but, of course, you can ignore that.

There does seem to be a problem with the generated man page and shell completions. It's deleting info related to a few commands. I think these were commands that @dduugg moved to the new command DSL recently so maybe that's related somehow. I really have no idea.

I ran the following and it changed some settings locally too but it tried to remove information related to a different command: brew bump-unversioned-casks

$ brew generate-man-completions
Writing completions to /usr/local/Homebrew/completions
Manpage and completions updated.
$ git diff --stats
 completions/bash/brew      | 24 ------------------------
 completions/fish/brew.fish | 14 --------------
 completions/zsh/_brew      | 19 -------------------
 docs/Manpage.md            | 21 ---------------------
 manpages/brew.1            | 14 --------------
 5 files changed, 92 deletions(-)
$ git log -1
commit 7b2bfee3638cba1571157f07fd447d9949bd0a6a (HEAD -> master, origin/master, origin/HEAD)
Merge: 0e981bd8f4 74aea8e92d
Author: Kevin <[email protected]>
Date:   Wed Mar 20 20:40:26 2024 -0700

    Merge pull request #16903 from apainintheneck/prevent-unexpected-network-calls-in-tests

    Prevent unexpected network calls in tests

Now I'm even more confused.

[MikeMcQuaid]

Thanks @razvanazamfirei!

[dduugg]

There does seem to be a problem with the generated man page and shell completions. It's deleting info related to a few commands. I think these were commands that @dduugg moved to the new command DSL recently so maybe that's related somehow. I really have no idea.

Thanks for flagging, i'll take a look.

[dduugg]

There does seem to be a problem with the generated man page and shell completions. It's deleting info related to a few commands. I think these were commands that @dduugg moved to the new command DSL recently so maybe that's related somehow. I really have no idea.

Thanks for flagging, i'll take a look.

Looks like I fat-fingered Homebrew::DevCmd::BumpUnversionedCask, should be Homebrew::DevCmd::BumpUnversionedCasks. I'll fix that and a test to validate classes map to their file/cmd names as expected.