Re: Pike 8.0: Sql.pgsql objects don't get garbage collected?

On Sat, Feb 27, 2016 at 10:03:13AM +0100, Stephen R. van den Berg wrote:
> Jeff Hungerford wrote:
> Did that imply that the "Attempting to clone an unfinished program"
> error cannot be induced anymore?  Because none of these last fixes
> should have affected that.

No, I am still seeing the "Attempting to clone an unfinished program" from
time to time at the beginning of a test run.

I'll try to come up with a test that can reliably trigger that error,

> >Indexing the NULL value with "i".
> >/usr/local/pike/8.0.181/lib/modules/Sql.pmod/pgsql.pike:630:
> >    pgsql(hungerf3@localhost:5432/hungerf3,0,0)->procmessage()
> 
> Well, maybe I eliminated this one too.  Small fix, checked in just now.

Looks like that one is fixed.  Thanks!

I just finished a long test run, and it looks like there are two more cases that
turn up:

This one happened multiple times:
-=-=-=-=-
Cannot call functions in destructed objects.
Unknown program: destructed object->function()
/usr/local/pike/8.0.181/lib/modules/Sql.pmod/pgsql.pike:624: unknown function(conxion  fd: 180 input queue: 0/224 queued portals: 1  output queue: 0/224 )
-=-=-=-=-

This one only turned up once across multiple long runs:
-=-=-=-=-
Lookup in destructed object.
/usr/local/pike/8.0.181/lib/modules/Sql.pmod/pgsql.pike:1265: unknown function()
-:1: Pike.Backend(1)->`()(4096.0)
/usr/local/pike/8.0.181/lib/modules/Sql.pmod/pgsql_util.pmod:83: Sql.pgsql_util.run_local_backend()
-=-=-=-=-

-Jeff