Open GoogleCodeExporter opened 9 years ago
It should also be noted that:
- The xcodebuild tool is completely no-functional when just the Command Line Tools are installed.
- Using xcode-select to set a bogus path results in xcodebuild hanging in an infinite loop rather than reporting to the user that it is non-functional.
Original comment by charles....@gmail.com
on 25 Aug 2012 at 8:07
Also, this can be re-produced on a non-fresh install by unsetting the
xcode-select path. This can be accomplished by moving or removing the contents
of `/usr/share/xcode-select/`.
Original comment by charles....@gmail.com
on 28 Aug 2012 at 5:22
Basically this bug means gyp depends on xcode when not all projects need it.
Now that Mac has released the Xcode developer tools, would be nice if Gyp would
respect that instead of blowing up if xcode-select doesn't exist.
Original comment by ankurset...@gmail.com
on 2 Feb 2013 at 4:13
Now that gyp has switched to using xcodebuild instead of xcode-select, the
workaround of setting the DEVELOPER_DIR environment variable no longer works at
all. The xcodebuild script that comes with the OS is just a shim, and the
Command Line Tools don't ship the real version of xcodebuild.
Hence - CLT-only users can't build any gyp projects ATM. This affects node, as
one example, which doesn't actually require Xcode to build but which fails to
configure under gyp.
Original comment by mistyde...@gmail.com
on 22 Apr 2013 at 6:26
Please fix this for CLT users.
Original comment by dylan.verheul
on 24 Apr 2013 at 4:58
Fix submitted: https://codereview.chromium.org/14887003
Original comment by mistyde...@gmail.com
on 3 May 2013 at 12:49
Could this patch please be accepted? This is still blocking node installs when
XCode is not present. As I mention on the original node.js ticket I was able to
cleanly work around the problem by applying this patch:
https://github.com/joyent/node/issues/3681
Original comment by d...@dbsgeo.com
on 23 Aug 2013 at 2:48
Anyone know the timeline of when this might be accepted or if it needs more
work?
Original comment by d...@dbsgeo.com
on 17 Sep 2013 at 1:58
Hi, I was able to verify that this patch fixed an issue I was having while
trying to compile a large OS project which had a submodule which uses gyp.
Would it be possible for someone from gyp to review this patch and/or consider
accepting it? Thanks! :)
Original comment by eric.sam...@gmail.com
on 7 Oct 2013 at 2:20
[deleted comment]
This is now fixed in r1819.
Original comment by mistyde...@gmail.com
on 16 Dec 2013 at 10:35
The patch was reverted, but has landed again in r1825. A second patch to fix
Xcode version detection is incoming.
Original comment by mistyde...@gmail.com
on 24 Dec 2013 at 10:08
Many of the patches to gyp mentioned above didn't work for me. I managed to get
this to work. In case any of you are still having problems, here's what I did:
I noticed that the source code for gyp is trying to run xcodebuild, but if
xcodebuild fails, then gyp fails. However, there is error handling in gyp in
case xcodebuild runs successfully, but returns something other than expected.
Relying on the behavior of gyp as described, I created an executable file
called xcodebuild and put it in my search path. Here are its contents:
#!/bin/bash
exit 0
That's it. All I wanted to do was make sure that gyp doesn't freak out when it
tries to run xcodebuild. I did not touch xcode-select or the gyp source.
This will probably break badly if there is a project that depends on Xcode,
though...
Original comment by James.G....@gmail.com
on 3 Feb 2014 at 4:12
Original issue reported on code.google.com by
charles....@gmail.com
on 25 Aug 2012 at 8:02