[Lldb-commits] [PATCH] Fix some posixy assumptions that were used during RunShellCommand

Zachary Turner zturner at google.com
Fri Dec 5 15:47:33 PST 2014


>>! In D6553#16, @clayborg wrote:
> Fine to commit as long as you don't set the architecture in the launch_info in Host::RunShellCommand() since that can break MacOSX.

Ok, I will take out the checks I added against the triple in ConvertArgumentsForLaunchingInShell for now and commit, then re-add them once the logic is properly separated in HostInfo.

Just so I can understand the code better, can you point me to the code that would break if the Architecture were set?  I see some in ConvertArgumentsForLaunchingInShell, but it's only exercised if will_debug == true, and RunShellCommand() uses will_debug == false.  So I'm assuming its something deep inside of Host::LaunchProcess.

http://reviews.llvm.org/D6553






More information about the lldb-commits mailing list