[Lldb-commits] [lldb] r358918 - [EditLineTest] Not always TERM is available, e.g. on some bots.

[Pavel Labath via lldb-commits]

On 22/04/2019 22:27, Davide Italiano via lldb-commits wrote:
> Author: davide
> Date: Mon Apr 22 13:27:10 2019
> New Revision: 358918
> 
> URL: http://llvm.org/viewvc/llvm-project?rev=358918&view=rev
> Log:
> [EditLineTest] Not always TERM is available, e.g. on some bots.
> 
> Modified:
>      lldb/trunk/unittests/Editline/EditlineTest.cpp
> 
> Modified: lldb/trunk/unittests/Editline/EditlineTest.cpp
> URL: http://llvm.org/viewvc/llvm-project/lldb/trunk/unittests/Editline/EditlineTest.cpp?rev=358918&r1=358917&r2=358918&view=diff
> ==============================================================================
> --- lldb/trunk/unittests/Editline/EditlineTest.cpp (original)
> +++ lldb/trunk/unittests/Editline/EditlineTest.cpp Mon Apr 22 13:27:10 2019
> @@ -244,17 +244,19 @@ private:
>     EditlineAdapter _el_adapter;
>     std::shared_ptr<std::thread> _sp_output_thread;
>   
> +protected:
> +  bool _has_term = true;
> +
>   public:
>     void SetUp() {
>       FileSystem::Initialize();
>   
>       // We need a TERM set properly for editline to work as expected.
> -    setenv("TERM", "vt100", 1);
> +    if (setenv("TERM", "vt100", 1) != 0)
> +      _has_term = false;

I'm confused as to what is this trying to solve. Judging by the manpage 
(which also agrees with my intuition) the setenv call can only fail 
under two conditions:
- "name (first arg) is NULL, points to a string of length 0, or contains 
an '=' character". This is clearly not the case here.
- "Insufficient memory to add a new variable to the environment". While 
technically possible, I find that highly unlikely, as you'd run into a 
lot more problems if the system was so severely memory-constrained.

I suspect this is only covering up some other problem. Can you elaborate 
on how you've arrived at this patch?

pl