On 2021-Mar-31, Tom Lane wrote:
> I wrote:
> > That is weird - only test 4 (of 8) runs at all, the rest seem to
> > fail to connect. What's different about pipelined_insert?
>
> Oh ... there's a pretty obvious theory. pipelined_insert is
> the only one that is not asked to write a trace file.
> So for some reason, opening the trace file fails.
> (I wonder why we don't see an error message for this though.)
.. oh, I think we forgot to set conn->Pfdebug = NULL when creating the
connection. So when we do PQtrace(), the first thing it does is
PQuntrace(), and then that tries to do fflush(conn->Pfdebug) ---> crash.
So this should fix it.
--
Álvaro Herrera Valdivia, Chile
<inflex> really, I see PHP as like a strange amalgamation of C, Perl, Shell
<crab> inflex: you know that "amalgam" means "mixture with mercury",
more or less, right?
<crab> i.e., "deadly poison"
diff --git a/src/interfaces/libpq/fe-connect.c b/src/interfaces/libpq/fe-connect.c
index a90bdb8ab6..56a8266bc3 100644
--- a/src/interfaces/libpq/fe-connect.c
+++ b/src/interfaces/libpq/fe-connect.c
@@ -3952,6 +3952,7 @@ makeEmptyPGconn(void)
conn->verbosity = PQERRORS_DEFAULT;
conn->show_context = PQSHOW_CONTEXT_ERRORS;
conn->sock = PGINVALID_SOCKET;
+ conn->Pfdebug = NULL;
/*
* We try to send at least 8K at a time, which is the usual size of pipe
