ne 4. 11. 2018 v 17:14 odesílatel Tom Lane <tgl@sss.pgh.pa.us> napsal:
Pavel Stehule <pavel.stehule@gmail.com> writes: > ne 4. 11. 2018 v 16:54 odesilatel Tom Lane <tgl@sss.pgh.pa.us> napsal: >> In short, I think it's a bug that we allow the above. If you >> want to keep the must-be-a-variable error then it should apply in >> this case too.
> I agree. This should be prohibited from PLpgSQL.
OK. In that case I'll run with my patch. The attached is what I had code-wise, but I've not changed the regression tests to match ... gotta go fix the docs too I guess.
I am not sure how safe is read argmodes from syscache after procedure execution. Theoretically, the procedure pg_proc tuple can be modified from procedure, and can be committed from procedure. Isn't better to safe argmodes before execution?