Re: check-functional skipUnless failure

["Philippe Mathieu-Daudé" <philmd@linaro.org>]

On 10/9/24 09:36, Thomas Huth wrote:
> On 10/09/2024 08.41, Philippe Mathieu-Daudé wrote:
>>> line 16, in tesseract_available
>>>      (stdout, stderr, ret) = run_cmd([ 'tesseract', '--version'])
>>>    File "/home/rth/qemu/src/tests/functional/qemu_test/cmd.py", line 
>>> 69, in run_cmd
>>>      subp = subprocess.Popen(args,
>>>    File "/usr/lib64/python3.9/subprocess.py", line 951, in __init__
>>>      self._execute_child(args, executable, preexec_fn, close_fds,
>>>    File "/usr/lib64/python3.9/subprocess.py", line 1837, in 
>>> _execute_child
>>>      raise child_exception_type(errno_num, err_msg, err_filename)
>>> FileNotFoundError: [Errno 2] No such file or directory: 'tesseract'


>> While looking at this, I'm surprised to notice has_cmd() actually
>> *runs* the command by calling run_cmd(). I'd have expected it solely
>> checks for binary presence and RX perms, avoiding unexpected side
>> effects.
> 
> It uses "which" to search for the binary ... that could be done better 
> for sure (especially since "which" might not be available on all 
> systems), but it already was done this way in the Avocado tests, so at 
> least that's not a regression.

Sure, I just wanted to share my surprise before forgetting ;)