Thread: isolationtester seems uselessly rigid as to length of permutation

I thought it'd be a good idea to put in some basic test cases for the
EvalPlanQual code using the isolationtester infrastructure.  While
fooling with it, I soon ran into this restriction:
    if (p->nsteps != nallsteps)    {        fprintf(stderr, "invalid number of steps in permutation %d\n", i + 1);
 exit_nicely();    }
 

ie, a "permutation" list has to specify exactly as many steps as there
are in the spec file.  This seems to me to be a useless restriction,
because it prevents running a test sequence that repeats some steps or
leaves other steps out.  Barring objections, I'm going to remove the
above lines.
        regards, tom lane

Excerpts from Tom Lane's message of sáb ene 28 18:08:36 -0300 2012:
> I thought it'd be a good idea to put in some basic test cases for the
> EvalPlanQual code using the isolationtester infrastructure.  While
> fooling with it, I soon ran into this restriction:
>
>         if (p->nsteps != nallsteps)
>         {
>             fprintf(stderr, "invalid number of steps in permutation %d\n", i + 1);
>             exit_nicely();
>         }
>
> ie, a "permutation" list has to specify exactly as many steps as there
> are in the spec file.  This seems to me to be a useless restriction,
> because it prevents running a test sequence that repeats some steps or
> leaves other steps out.  Barring objections, I'm going to remove the
> above lines.

Yes, sorry, that patch was a mistake (misdiagnosed problem, later
patched differently).  My FOR KEY SHARE patch deals with that too, but
I'm happy with just removing the test.

--
Álvaro Herrera <alvherre@commandprompt.com>
The PostgreSQL Company - Command Prompt, Inc.
PostgreSQL Replication, Consulting, Custom Development, 24x7 support