• [ X] Have you followed the guidelines in our Contributing document?
• [ X] Have you checked to ensure there aren't other open Pull Requests for the same change?
• [ X] 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.
• [X ] Have you successfully run brew style with your changes locally?
• [ ] Have you successfully run brew typecheck with your changes locally?
• [ X] Have you successfully run brew tests with your changes locally?

It keeps happening. Maybe it's because of WSL?

All I wanted was an easy way to install AVR-GCC on Ubuntu. If you're talking about the issue you encountered in osx-cross/homebrew-avr#230, wouldn't it have been easier to wrap args << "SED=/usr/bin/sed" in an on_macos block? I suppose getting rid of shims that are no longer necessary is a nice bonus...

Yes, that would have been the easy and wrong solution, that'd only proliferate the confusion people end up in over these shims. :)

All I wanted was an easy way to install AVR-GCC on Ubuntu.

If you're talking about the issue you encountered in https://github.com/osx-cross/homebrew-avr/issues/230, wouldn't it have been easier to wrap args << "SED=/usr/bin/sed" in an on_macos block?

I suppose getting rid of shims that are no longer necessary is a nice bonus...

• [x] Have you followed the guidelines for contributing?
• [x] Have you checked that there aren't other open pull requests for the same formula update/change?
• [ ] Have you built your formula locally with brew install --build-from-source <formula>, where <formula> is the name of the formula you're submitting?
• [ ] Is your test running fine brew test <formula>, where <formula> is the name of the formula you're submitting?
• [ ] Does your build pass brew audit --strict <formula> (after doing brew install <formula>)?
• [ ] Have you included the output of brew gist-logs <formula> of the build failure if your PR fixes a build failure. Please quote the exact error message.

Waiting on Homebrew/brew#10802 to be merged.

I'm expecting checks to fail for now with

Error: Files were found with references to the Homebrew shims directory.

hence this PR is still in draft. Once the above is merged I'll trigger another set of checks.

OK.

This is for apr-1-config.

Oops, didn't catch that bit! I see it now. I've fallen deep enough down this rabbit hole; that can be someone else's problem. :)

I'll try to submit PRs over at homebrew-core and linuxbrew-core too to clean up any sed workarounds later.

All I wanted was an easy way to install AVR-GCC on Ubuntu. I haven't even opened my ~2011 MacBook Air for maybe 6 months. :)

I'm confused, is this not a bug then, given it was closed? With Python I can do brew link --overwrite [email protected] to make it my default and then brew link --overwrite python to go back to the latest. Why would I not be able to do this with openjdk?

Also a separate question, every time a newer version of Python is updated (e.g. 3.9) it re-links against that version so I have to overwrite again. Is there any better concept of "default" that doesn't constantly get reset with each upgrade of a newer version of that package?

couldn't find any logic that handles an s flag

The sed shim you deleted handled this flag: if [[ "$HOMEBREW_CCCFG" = *s* ]]

I don't see any for the a flag either here, so should that go too?

This is for apr-1-config. While this no longer exists on modern macOS, I think it does on Yosemite (and some later versions - not sure exactly when it disappeared). I'd like someone to test the tool on Yoesmite before we remove it, as the bug may still be relevant.

The sed issue was never really seen as a "bug" by Apple and it was more of a backwards compatibility issue at the time.

I'm happy to try this, but these two lines should be removed too:

Will include your suggestion @Bo98. I did a quick git grep HOMEBREW_CCCFG but couldn't find any logic that handles an s flag, so I guess that's been superfluous code for some time? I don't see any for the a flag either here, so should that go too?

I've just noticed a whole bunch of workarounds for the sed shim in homebrew-core :(

brew gist-logs <formula> link OR brew config AND brew doctor output

> brew config
HOMEBREW_VERSION: 3.0.4
ORIGIN: https://github.com/Homebrew/brew
HEAD: ee52b19179d056f3dd2fdf9c52214d3d51dbb9d9
Last commit: 4 days ago
Core tap ORIGIN: https://github.com/Homebrew/linuxbrew-core
Core tap HEAD: 145f4964db7896156b1bcbcf5181b7e5f6a638a8
Core tap last commit: 7 minutes ago
Core tap branch: master
HOMEBREW_PREFIX: /home/linuxbrew/.linuxbrew
HOMEBREW_CASK_OPTS: []
HOMEBREW_DISPLAY: :0
HOMEBREW_MAKE_JOBS: 24
Homebrew Ruby: 2.6.3 => /home/linuxbrew/.linuxbrew/Homebrew/Library/Homebrew/vendor/portable-ruby/2.6.3_2/bin/ruby
CPU: 24-core 64-bit zen
Clang: N/A
Git: 2.25.1 => /usr/bin/git
Curl: 7.68.0 => /usr/bin/curl
Kernel: Linux 5.4.0-45-generic x86_64 GNU/Linux
OS: Linux Mint 20 (ulyana)
Host glibc: 2.31
/usr/bin/gcc: 9.3.0
/usr/bin/ruby: 2.7.0
glibc: N/A
gcc@5: N/A
xorg: N/A

> brew doctor
Your system is ready to brew.

• [X] The brew doctor above contains no "Warning" lines.

What were you trying to do (and why)?

brew install argocd

What happened (include all command output)?

> brew install argocd
==> Auto-updated Homebrew!
Updated 1 tap (homebrew/core).
==> Updated Formulae
Updated 1 formula.

==> Downloading https://linuxbrew.bintray.com/bottles/go-1.16_1.x86_64_linux.bottle.tar.gz
Already downloaded: /home/sdelrio/.cache/Homebrew/downloads/303d221902a438336a16ccf086eac2b5920490516545955e7552cc8667f96190--go-1.16_1.x86_64_linux.bottle.tar.gz
==> Cloning https://github.com/argoproj/argo-cd.git
Updating /home/sdelrio/.cache/Homebrew/argocd--git
==> Checking out tag v1.8.7
HEAD is now at eb3d1fb Bump version to 1.8.7
HEAD is now at eb3d1fb Bump version to 1.8.7
Error: v1.8.7 tag should be a97230859344f8906502763743717d7ad7339c55
but is actually eb3d1fb84b9b77cdffd70b14c4f949f1c64a9416

What did you expect to happen?

argocd 1.8.7 installed

Step-by-step reproduction instructions (by running brew commands)

brew install argocd

as commented on https://github.com/Homebrew/linuxbrew-core/pull/22480 opened issue + PR on homebrew-core

• https://github.com/Homebrew/homebrew-core/issues/72633
• https://github.com/Homebrew/homebrew-core/pull/72635

I'd rather this specifically checked the Formula directory.

Currently things like lib/test_bot.rb also get picked up.

I'm happy to try this, but these two lines should be removed too: https://github.com/Homebrew/brew/blob/a65c5d685c1988007882bdbb562799df6e1d4f3d/Library/Homebrew/extend/os/mac/extend/ENV/super.rb#L101-L102

We could always use gcc-10 for these cases, I agree, but it has two downsides:

• When we migrate from Ubuntu 16.04 to Ubuntu 18.04 for CI (which is the next big project I would like to start soon), we will have to downgrade all the formulae that depend on gcc-10 to system gcc-7 (if it builds with systems gcc-7).
• Some formulae will probably not build with gcc-10 because it's too new: we still will need to go back to gcc-7/8/9 for some of them.

The Mac patch was added here: https://github.com/Homebrew/formula-patches/commit/4dcf52867baa32f038fa2ab1285fd1364eb65846

Both patches do almost the same, but would need to be merged together and upstreamed.

Upstream's last change is from July 2020: https://sourceforge.net/p/freeimage/svn/HEAD/tree/, it looks like a smaller project, with not that much activity. I do not have much time to invest in this right now.

I definitively are in favour of reducing the number of patches and be more strict about them: I can start that project in 2022 when I'm done with the linuxbrew-core merge :D

One important question that will affect how we resolve this is whether we want to always force the latest version of gcc if gcc-5 (or gcc-7 in the future) is not compatible.

If we went down this path, it would be simpler from the perspective of dependencies (we wouldn't have to keep track of the minimum version of GCC that works for each formula), but would mean forcing the user to install brewed GCC even if they have a new enough system GCC.

I am leaning in favor of this for a couple of reasons. When we have fully relocatable binaries, users should rarely if ever be compiling from source. Even in that scenario or for users who have /home/linuxbrew/.linuxbrew as their prefix, we still have to make sure the system GCC runtime libs are new enough to be compatible with whatever we built with in CI.

So for example if we used gcc-8 to build in CI, we'd have to make sure that the user's system GCC was at least gcc-8 - otherwise they would need brewed gcc-8 installed to give them new enough runtime libs.

However, if we just always build anything that needs a newer GCC than gcc-5 with gcc-10, then we can always use brewed gcc-10 and be sure that we will have new enough runtime libs. Although this is technically "wasteful" in the sense that a user who system GCC is gcc-10 doesn't really need brewed gcc-10, for all the users who have an older GCC, this would be better than having 5 different versions GCC installed.

Mountain Lion isn't even supported by Homebrew anymore.

This shim hails back to Homebrew/legacy-homebrew#13787 @MikeMcQuaid

This gets picked up by configure scripts (which is fine during builds), but occasionally gets baked into binaries, triggering:

Error: Files were found with references to the Homebrew shims directory.

This in turn causes tap maintainers to just override SED=/usr/bin/sed for configure, which doesn't always work properly: osx-cross/homebrew-avr#230

Other instances:

• https://stackoverflow.com/questions/40357246/usr-local-library-homebrew-shims-super-sed-no-such-file-or-directory
• https://github.com/laruence/yaconf/issues/17
• https://github.com/Amar1729/homebrew-formulae/issues/1
• https://blog.logical-dice.com/articles/wp/295

• [x] Have you followed the guidelines in our Contributing document?
• [x] Have you checked to ensure there aren't other open Pull Requests for the same change?
• [x] 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?

I have opened https://github.com/Homebrew/homebrew-core/pull/72632

Merge Homebrew/homebrew-core into Homebrew/linuxbrew-core

• [ ] p7zip

Merge Homebrew/homebrew-core into Homebrew/linuxbrew-core

• [ ] p7zip

• [x] Have you followed the guidelines for contributing?
• [x] Have you checked that there aren't other open pull requests for the same formula update/change?
• [ ] Have you built your formula locally with brew install --build-from-source <formula>, where <formula> is the name of the formula you're submitting?
• [ ] Is your test running fine brew test <formula>, where <formula> is the name of the formula you're submitting?
• [ ] Does your build pass brew audit --strict <formula> (after doing brew install <formula>)?
• [ ] Have you included the output of brew gist-logs <formula> of the build failure if your PR fixes a build failure. Please quote the exact error message.