Re: [HACKERS] psql - add special variable to reflect the last querystatus - Mailing list pgsql-hackers

From Fabien COELHO
Subject Re: [HACKERS] psql - add special variable to reflect the last querystatus
Date
Msg-id alpine.DEB.2.20.1706280951170.1763@lancre
Whole thread Raw
In response to Re: [HACKERS] psql - add special variable to reflect the last query status  (Pavel Stehule <pavel.stehule@gmail.com>)
Responses Re: [HACKERS] psql - add special variable to reflect the last query status  (Pavel Stehule <pavel.stehule@gmail.com>)
List pgsql-hackers
Hello Pavel,

> + if (success)
> + {
> + char   *ntuples = PQcmdTuples(results);
> + SetVariable(pset.vars, "ROW_COUNT", *ntuples ? ntuples : "0");
> + SetVariable(pset.vars, "ERROR", "FALSE");
> + }
> + else
> + {
> + SetVariable(pset.vars, "ROW_COUNT", "0");
> + SetVariable(pset.vars, "ERROR", "TRUE");
> + }
> +}
>
> It can be simplified
>
> SetVariable(pset.vars, "ROW_COUNT", success ? PQcmdTuples(results) : 0);

According to the documentation, PQcmdTuples returns "" in some cases and 
ISTM we want "0" instead for consistency, so that it is always a number. I 
rejected calling PQcmdTuples twice:
  ..., success && *PQcmdTuples(results) ? PQcmdTuples(results) : "0")

Thus it makes the "if (success)" necessary for ROW_COUNT, and then it 
looked simpler to handle ERROR the same way.

Now if the semantics is changed to put as row count whatever comes out of 
the function, even if not a count, then the code could indeed be 
simplified as you suggest.

-- 
Fabien.



pgsql-hackers by date:

Previous
From: Pavel Stehule
Date:
Subject: Re: [HACKERS] psql - add special variable to reflect the last query status
Next
From: Pavel Stehule
Date:
Subject: Re: [HACKERS] psql - add special variable to reflect the last query status