superenv: help Autotools with 10.12 SDK on 10.11 #970

ilovezfs · 2016-09-15T09:39:23Z

The GNU Autotools tests for whether a given symbol is defined are
reliably coming to incorrect conclusions on 10.11 with the 10.12 SDK
in Xcode 8. This overrides its decisions by forcing the right answer
in superenv using ac_cv_func_* environment variables and setting them to
"no" on 10.11. The list of problematic symbols is from

grep 'weak$os10.11' MacOSX.sdk/usr/lib/system/libsystem_c.tbd

MikeMcQuaid

Makes sense to me. Great work on this 👏

MikeMcQuaid · 2016-09-15T09:42:55Z

Library/Homebrew/extend/os/mac/extend/ENV/super.rb

+    # Filter out symbols known not to be defined on 10.11 since GNU Autotools
+    # can't reliably figure this out with Xcode 8 on its own yet. The Xcode
+    # version doesn't need to be checked since these are harmless with Xcode 7
+    # even if they're unnecessary.


Personally I'd add the Xcode check anyway because better safe than sorry but will leave that decision up to you.

if MacOS.version == "10.11" && DevelopmentTools.clang_build_version >= 800

yes?

Yeh, or you can do MacOS::Xcode.version >= "8.0"

My concern with that is that it won't fix the problem on a CLT-only install assuming Apple puts the same stuff in the 10.11 Xcode 8 CLT (if there is one actually coming out).

Ah, I see. Yes, your version is better then. CC @DomT4 to confirm about the CLT?

I'm not sure about the CLT on 10.11 yet to be honest because Apple releases CLT packages scoped to individual platforms, i.e. there's one available for macOS Sierra but not El Capitan, but when El Capitan gets its version it'll get its own .pkg.

It seems feasible they could push a CLT for 10.11 without the problems, but who knows until they do.

let's avoid speculatively fixing bugs and just do

if MacOS.version == "10.11" && MacOS::Xcode.installed? && MacOS::Xcode.version.to_i >= 8

We can always expand its scope later as needed.

Agreed except avoid the .to_i and just do MacOS::Xcode.version >= "8.0" 😉

Shall I change https://github.com/ilovezfs/brew/blob/2505da3182f48c960e896f42173b4d9809e01001/Library/Homebrew/extend/os/mac/extend/ENV/super.rb#L92 as well? hehe

Yeh, feel free 😁

zmwangx

I wouldn't have thought OS X doesn't have basename_r and dirname_r lol.

DomT4

Part of me wonders if it's better to add this change as an unique method like determine_cccfg, etc. I can't imagine this problem getting smaller in future years, and it may be easier to keep track of that way.

No big preference either way though, just a thought. Thanks for working on this & getting a PR pushed so quickly 🙇.

The GNU Autotools tests for whether a given symbol is defined are reliably coming to incorrect conclusions on 10.11 with the 10.12 SDK in Xcode 8. This overrides its decisions by forcing the right answer in superenv using ac_cv_func_* environment variables and setting them to "no" on 10.11. The list of problematic symbols is from grep 'weak$os10.11' MacOSX.sdk/usr/lib/system/libsystem_c.tbd

ilovezfs · 2016-09-15T10:41:55Z

PR refreshed.

Part of me wonders if it's better to add this change as an unique method like determine_cccfg, etc. I can't imagine this problem getting smaller in future years, and it may be easier to keep track of that way.

It may eventually benefit from something like that, but I think for now that's probably over-engineering it.

mistydemeo

🙇

DomT4 · 2016-09-15T11:26:45Z

👍. I'm happy unless you're planning to add a test for this change or anything. Otherwise, let's ship it and see what breaks.

ilovezfs · 2016-09-15T11:35:34Z

🚀

DomT4 · 2016-09-15T11:35:35Z

Can confirm this at least fixes xz's runtime failure of false empty archive messages & installing python on 10.11 with Xcode 8.

People will need to reinstall xz and other impacted-at-run formulae by the looks of it.

ilovezfs · 2016-09-15T11:37:21Z

People will need to reinstall xz and other impacted-at-run formulae by the looks of it.

indeed.

chdiza · 2016-09-15T14:41:06Z

I got a malfunctioning xz with Xcode 8 on Sierra, which also was fixed by this PR (as far as I can tell).

chdiza · 2016-09-15T15:17:08Z

That is to say: even with Xcode 8 and the associated CLT, on Sierra, xz gave me the "empty archive" nonsense and any formula I built whose tarball was an xz-ball was defective ("illegal instruction 4").

But after this PR it seemed to work.

zmwangx · 2016-09-15T15:18:11Z

This PR specifically has a MacOS.version == "10.11" guard, so I doubt it should have any effect on your Sierra environment.

ilovezfs · 2016-09-15T15:19:31Z

@chdiza what's brew config have to say for itself?

chdiza · 2016-09-16T02:12:20Z

brew --config has this to say:

HOMEBREW_VERSION: 0.9.9
ORIGIN: https://github.com/Homebrew/brew
HEAD: a148aa3a41d34a07caab5d17daa26b4749ff3c50
Last commit: 15 hours ago
Core tap ORIGIN: https://github.com/Homebrew/homebrew-core
Core tap HEAD: 0e99665f347c1554f55fbecf448956a44ccd99e5
Core tap last commit: 14 hours ago
HOMEBREW_PREFIX: /usr/local
HOMEBREW_REPOSITORY: /usr/local/hb
HOMEBREW_CELLAR: /usr/local/Cellar
HOMEBREW_BOTTLE_DOMAIN: https://homebrew.bintray.com
CPU: quad-core 64-bit ivybridge
Homebrew Ruby: 2.0.0-p648
Clang: 8.0 build 800
Git: 2.8.4 => /Applications/Xcode.app/Contents/Developer/usr/bin/git
Perl: /usr/bin/perl
Python: /usr/bin/python
Ruby: /usr/bin/ruby => /System/Library/Frameworks/Ruby.framework/Versions/2.0/usr/bin/ruby
Java: N/A
OS X: 10.12-x86_64
Xcode: 8.0
CLT: 8.0.0.0.1.1472435881
X11: N/A

Maybe it wasn't this particular commit that fixed the problem, but another one that landed within around 24 hours.

Erlang uses some customized Autotools tests, so the general superenv workaround in Homebrew/brew#970 is insufficient

Erlang uses some customized Autotools tests, so the general superenv workaround in Homebrew/brew#970 is insufficient Closes #4920. Signed-off-by: ilovezfs <[email protected]>

gnutls uses some customized Autotools tests, so the general superenv workaround in Homebrew/brew#970 is insufficient

BrewTestBot added the in progress Maintainers are working on this label Sep 15, 2016

MikeMcQuaid approved these changes Sep 15, 2016

View reviewed changes

MikeMcQuaid reviewed Sep 15, 2016

View reviewed changes

zmwangx approved these changes Sep 15, 2016

View reviewed changes

DomT4 approved these changes Sep 15, 2016

View reviewed changes

zmwangx mentioned this pull request Sep 15, 2016

"Empty archive" error when install formulas with tar.xz archives, after upgrading to Xcode 8 #969

Closed

ilovezfs force-pushed the ac_cv_xcode8 branch from 2505da3 to 1d7aa1f Compare September 15, 2016 10:37

mistydemeo approved these changes Sep 15, 2016

View reviewed changes

ilovezfs merged commit a148aa3 into Homebrew:master Sep 15, 2016

BrewTestBot removed the in progress Maintainers are working on this label Sep 15, 2016

DomT4 mentioned this pull request Sep 15, 2016

wine failed to build on 10.11.6 Error: Empty archive Homebrew/homebrew-core#4820

Closed

zmwangx mentioned this pull request Sep 15, 2016

OS X v10.11.x 'El Capitan,' Xcode 8, and the Command Line Tools (Or, Rather, the Lack Thereof…) (Tracking) #972

Closed

3 tasks

ilovezfs added a commit to ilovezfs/homebrew-core that referenced this pull request Sep 17, 2016

erlang: fix the build on 10.11 with Xcode 8

74fed27

Erlang uses some customized Autotools tests, so the general superenv workaround in Homebrew/brew#970 is insufficient

ilovezfs mentioned this pull request Sep 17, 2016

erlang: fix the build on 10.11 with Xcode 8 Homebrew/homebrew-core#4920

Closed

zmwangx mentioned this pull request Sep 17, 2016

[ffmpeg.rb] dyld: lazy symbol binding failed: Symbol not found: _clock_gettime Homebrew/homebrew-core#4902

Closed

3 tasks

ilovezfs added a commit to ilovezfs/homebrew-core that referenced this pull request Oct 18, 2016

gnutls: fix the build on 10.11 with Xcode 8

ba36c36

gnutls uses some customized Autotools tests, so the general superenv workaround in Homebrew/brew#970 is insufficient

ilovezfs mentioned this pull request Oct 18, 2016

gnutls: fix the build on 10.11 with Xcode 8 Homebrew/homebrew-core#6041

Merged

ilovezfs added a commit to Homebrew/homebrew-core that referenced this pull request Oct 18, 2016

gnutls: fix the build on 10.11 with Xcode 8 (#6041)

af2c4dd

gnutls uses some customized Autotools tests, so the general superenv workaround in Homebrew/brew#970 is insufficient

Homebrew locked and limited conversation to collaborators May 3, 2018

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

superenv: help Autotools with 10.12 SDK on 10.11 #970

superenv: help Autotools with 10.12 SDK on 10.11 #970

ilovezfs commented Sep 15, 2016

MikeMcQuaid left a comment

MikeMcQuaid Sep 15, 2016

ilovezfs Sep 15, 2016

MikeMcQuaid Sep 15, 2016

ilovezfs Sep 15, 2016

MikeMcQuaid Sep 15, 2016

DomT4 Sep 15, 2016

ilovezfs Sep 15, 2016

MikeMcQuaid Sep 15, 2016

ilovezfs Sep 15, 2016

MikeMcQuaid Sep 15, 2016

zmwangx left a comment

DomT4 left a comment

ilovezfs commented Sep 15, 2016

mistydemeo left a comment

DomT4 commented Sep 15, 2016

ilovezfs commented Sep 15, 2016

DomT4 commented Sep 15, 2016

ilovezfs commented Sep 15, 2016

chdiza commented Sep 15, 2016

chdiza commented Sep 15, 2016

zmwangx commented Sep 15, 2016

ilovezfs commented Sep 15, 2016

chdiza commented Sep 16, 2016

superenv: help Autotools with 10.12 SDK on 10.11 #970

superenv: help Autotools with 10.12 SDK on 10.11 #970

Conversation

ilovezfs commented Sep 15, 2016

MikeMcQuaid left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

zmwangx left a comment

Choose a reason for hiding this comment

DomT4 left a comment

Choose a reason for hiding this comment

ilovezfs commented Sep 15, 2016

mistydemeo left a comment

Choose a reason for hiding this comment

DomT4 commented Sep 15, 2016

ilovezfs commented Sep 15, 2016

DomT4 commented Sep 15, 2016

ilovezfs commented Sep 15, 2016

chdiza commented Sep 15, 2016

chdiza commented Sep 15, 2016

zmwangx commented Sep 15, 2016

ilovezfs commented Sep 15, 2016

chdiza commented Sep 16, 2016