On 2024-07-02 16:21 +0200, Tom Lane wrote:
> Erik Wienhold <ewie@ewie.name> writes:
> > On 2024-07-02 10:45 +0200, Peter Eisentraut wrote:
> >> I had to revert the test case from the 0002 patch. It ended up running some
> >> build farm machines out of memory.
>
> >> +ERROR: out of memory
> >> +DETAIL: Failed on request of size 1073741820 in memory context "PortalContext".
>
> > That means paramno is less than MaxAllocSize/sizeof(Oid) if it tries to
> > allocate memory. MaxAllocSize is always 0x3fffffff. Is sizeof(Oid)
> > less than 4 on those machines?
>
> No. Y'know, it's not really *that* astonishing for a machine to not
> have a spare 1GB of RAM available on-demand. This test would
> certainly have failed on our 32-bit animals, although it doesn't
> look like any of them had gotten to it yet.
Ah, sorry. I somehow missed that it allocates memory for each param,
instead of first checking *all* params. m(
--
Erik
