Sierra indigo issue73

[lloydc99]

Include private version of qt4 for Sierra since it is not provided by main brew.

Patch bug in include headers in ros_console until fix is moved back to indigo.

Add a link to nosetests because it does not seem to be in the brew path.

Change the default search order for headers to look for OSX Frameworks last to get around issue #73.

[mikepurvis]

Wow, thanks for sorting this all out.

The CMAKE_FIND_FRAMEWORK thing is creepy to me; can you explain more about what's going on with that?

Also, the travis jobs are now all failing due to the long build time associated with the qt4 tap. I wonder if there's a way we can bottle that to avoid the long wait?

[mikepurvis]

My preference would be to handle this via a rosdep override, if possible.

@wjwwood Is rosdep able to tell apart different versions of OS X the way it can different versions of Ubuntu?

[lloydc99]

I'm guessing you mean putting the brew in a rosdep.
I think I still need the SKIP_KEYS.

Would you want to switch the rosdep override based on osx version here in the script?

[lloydc99]

I totally agree on the creepy factor. I found it by grepping through the cmake modules.
It seemed related to the fix I found in removing the System Framework include in the minimal example.

I don't know much more than the doc page.
https://cmake.org/cmake/help/v3.4/variable/CMAKE_FIND_FRAMEWORK.html

As far as bottling qt4, I'm pretty much a brew noob and ruby rookie. I can give it a shot, but it might take a while.

[mikepurvis]

Based on the Travis build, it looks like El Cap is also using the forked qt4 formula— wasn't the intention that only Sierra would use it, and that the others could stick with upstream (desirable in part because then it's upstream taking care of bottling).

[lloydc99]

I'll work on the conditional. Hopefully that is the problem.
Glad there is a test machine!

[mikepurvis]

There's a Travis image for xcode 8.2— that might be a Sierra image, if you want to add it to the travis config and get some coverage of this new logic:

https://docs.travis-ci.com/user/osx-ci-environment/#Xcode-version

Looks like xcode 7 should also be moved to xcode 7.3.1.

[lloydc99]

I've got two of the travis tests to pass.

The 7.3 failure looks like a time-out because the sw_vers string being returned is not 10.10.x but 10.11.6. There is no bottle in osrf/simulation/simbody for El Capitan or Sierra.

Tapped 20 formulae (68 files, 101.1K)
OSX Product version:  10.11.6
Installing custom El Capitan/Sierra qt4 with pyside.

The 8.2 full_desktop build fails for some strange conflict over f2py while installing numpy. I saw this on my El Capitan machine at work and had to manually delete the file. No idea how it got there.
I'm also not sure why is is affecting matplotlib.

[mikepurvis]

Thanks for all the hard work here, however via #84 we now have a setup that works with QT5, so this is unnecessary.

[lloydc99]

Yeah, I see a bunch of stuff moving again. Sadly for my Baxter I need to get indigo working again.

…

On Fri, Oct 20, 2017 at 3:13 PM, Mike Purvis ***@***.***> wrote: Thanks for all the hard work here, however via #84 <#84> we now have a setup that works with QT5, so this is unnecessary. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#74 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AF103JxJOzyxTIAaV-6YUk_EjQRKkerqks5suPD1gaJpZM4LDMFL> .

[mikepurvis]

What is it you need specifically from Indigo? Can you not just build Baxter's indigo packages from source against the Kinetic/Lunar base?

[lloydc99]

I had all of the Rethink SDK working last November. I really had not considered moving the SDK to lunar. I figured it would be too hard. Getting ROS to build on OSX has caused me to redefine "too hard." I may give that a try first. Now I have to get past the highSierra matplotlib issue.

…

On Fri, Oct 20, 2017 at 9:07 PM, Mike Purvis ***@***.***> wrote: What is it you need specifically from Indigo? Can you not just build Baxter's indigo packages from source against the Kinetic/Lunar base? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#74 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AF103CQXYzuhCJwZS4Ho6pb8EnuF572Mks5suUPJgaJpZM4LDMFL> .

[jmtatsch]

@lloydc99 matplotlib issue in high sierra is already solved here https://github.com/Homebrew/homebrew-science/pull/6382