envoy 1.19.0

[codefromthecrypt]

Signed-off-by: Adrian Cole adrian@tetrate.io

• Have you followed the guidelines for contributing?
• Have you ensured that your commits follow the commit style guide?
• 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 --build-from-source <formula>)? If this is a new formula, does it pass brew audit --new <formula>?

---

[carlocab]

Please split your commits so that each commit modifies at most one formula. Thanks!

I'm also not certain about adding another versioned Envoy formula. One is fine, but two is probably a bit much, especially given that it works on only two out of five supported platforms.

[codefromthecrypt]

arm isn't expected to work as nothing upstream improved on this see #79312 and related

[codefromthecrypt]

@carlocab sounds good. so I will replace 1.17 with 1.18 that's fine with me.

[carlocab]

Looks like Envoy failed to build with the same error as in the previous PR. #81266

[codefromthecrypt]

cool was waiting for a failed build on a branch with write access. likely Xcode related per some investigations from @mathetake

[codefromthecrypt]

ps any clue on "The homepage URL https://www.envoyproxy.io is not reachable" could it be timeout related?

[codefromthecrypt]

to fix the audit issue is easier, basically Envoy's Netlify is redirecting unless you hit something. Adding index.html will sort it out.

[codefromthecrypt]

the problem is upstream compatibility with xcode 12 envoyproxy/envoy#17393

suppose we can figure out how to revert to an earlier version for these formulas as ^^ won't be in the source history until a later version of envoy

[codefromthecrypt]

this isn't needed as it doesn't run envoy's unit tests

[codefromthecrypt]

this works around the xcode issue until it is fixed upstream

[codefromthecrypt]

thanks especially to @mathetake as he helped form the syntax. I'm not knowledgeable on this part.

[codefromthecrypt]

looks like an impossible combination.. we need to downgrade or delete this line https://github.com/envoyproxy/envoy/blob/main/bazel/envoy_internal.bzl#L69

[codefromthecrypt]

watching paint dry to test locally as the build takes so long, but here's my idea for patching the file

diff --git a/Formula/envoy.rb b/Formula/envoy.rb
index 471b5c5099..38887f2ef8 100644
--- a/Formula/envoy.rb
+++ b/Formula/envoy.rb
@@ -22,6 +22,8 @@ class Envoy < Formula
   depends_on macos: :catalina
 
   def install
+    # Work around xcode 12 incompatibility until envoyproxy/envoy#17393
+    inreplace "bazel/envoy_internal.bzl", '"-Wrange-loop-analysis",', ""
     args = %W[
       -c
       opt

[codefromthecrypt]

at least in the build dir the above replacement worked locally even if the build won't finish for a few hours. I pushed here jic it finishes fine

[codefromthecrypt]

trying both deleting the explicit entry for the linter from the config file and also adding its exclude as an arg. Maybe it is implicitly on by default. If this doesn't work and still no progress upstream...

maybe we temporarily downgrade to the previous version of xcode. this project is extremely complex in terms of flags and attempts cost 3-5hrs each depending on how fast your machine is, and with the build is running it takes it completely over

[codefromthecrypt]

envoyproxy/envoy#17428 is the current approach to solve this

[codefromthecrypt]

ok switched approach to an inline patch

[codefromthecrypt]

I'm not sure the world record on slowest build, but one dimension is fast approaching 8hrs. 🍿

[SMillerDev]

Unfortunately that one is taken by rust/go/python builds. Those often are over 24 hours 😅 .

[codefromthecrypt]

one can dream..

[codefromthecrypt]

Successful in 481m!

ok this is now passing more than before. There's a patch aka tech debt, but frankly the upstream patch...

1. isn't merged so would only effect a later version
2. isn't guaranteed to be back ported

and regardless I'm on the hook at least for a while, so don't mind playing revert duty if/when above change

[carlocab]

We can use a patch from the upstream PR instead of carrying it around in the formula, I think.

[codefromthecrypt]

thanks for the help as usual @carlocab I will re-organize the commits per formula again and 🤞

[codefromthecrypt]

ok finally green!

[carlocab]

Oops; this got buried under a bunch of other PRs. LGTM. Thanks, @codefromthecrypt!