The max. length of runner_cfg is defined by runner_cfg_default length (Xcelium)

[VinieBerry]

In many VHDL examples such as check the runner_cfg generic has the runner_cfg_default default value. This generic value includes the simulation output path based on the run.py file location (vunit_cli.py).

However, during simulation I get the following error (Xcelium 21.09):

cannot open '/work/sesavunit_results' for writing (Permission denied)

          File: /work/sesa595203/vunit/vunit/vhdl/core/src/core_pkg.vhd, line = 32, pos = 48
         Scope: @vunit_lib.core_pkg:setup
          Time: 0 FS + 0

Yet, the arguments provided to the simulators are okay:

-generic "tb_example.runner_cfg => \"active python runner : true,enabled_test_cases : ,output path : /work/sesa595203/vunit/examples/vhdl/check/vunit_out/test_output/lib.tb_example.all_7b5933c73ddb812488c059080644e9fd58c418d9/,tb path : /work/sesa595203/vunit/examples/vhdl/check/,use_color : true\""

Actually, the issue seems that the runner_cfg string length is constrained by the runner_cfg_default string length, itself defined by its content set in run_types.vhd file (74 chars) :

constant runner_cfg_default : string := "enabled_test_cases : __all__, output path : , active python runner : false";

Therefore, the runner_cfg value is cropped:

active python runner : true,enabled_test_cases : ,output path : /work/sesa

How should we deal with this issue ?

[std-max]

With which VHDL standard do you compile ?

[LarsAsplund]

The default value supports the standalone mode to run a testbench without a Python runner. That's very much a corner case today so we could really remove the default from the examples.

However, there are coming features (#772) which adds other defaults so in general this needs fixing. We could add an end marker like NUL to string and then pad with additional characters to extend the length. The problem is to figure out how long is long enough. I recall that there was a simulator that had a limit in the past. Not sure what the status is today.