[nexmark] Invoke run.py via python interpreter.

Leonid Ryzhyk · ryzhyk · commit a1a2830a7bd4 · 2024-10-08T01:39:52.000Z
This small fix should make the benchmark script more portable. The
script invoked `run.py` as an executable, relying on the
`#!/usr/bin/python3` annotation in that file to pick the interpreter for
it. This didn't work on my system using a virtual environment, which
requires running the python interpreter from the virtual environment,
instead of the one in `/usr/bin`.

The patch invokes run.py via `python3 run.py`, which should pick the
python interpreter from the virtual environment if one is activated.

Signed-off-by: Leonid Ryzhyk &lt;leonid@feldera.com&gt;
diff --git a/benchmark/run-nexmark.sh b/benchmark/run-nexmark.sh
@@ -477,7 +477,7 @@ case $runner:$language in
 
     feldera:sql)
         rm -f results.csv
-        PYTHONUNBUFFERED=1 run_log feldera-sql/run.py \
+        PYTHONUNBUFFERED=1 run_log python3 feldera-sql/run.py \
             --api-url="$api_url" \
             --cores $cores \
             --events $events \
