On 2025-09-26 Fr 8:06 PM, Jacob Champion wrote:
On Fri, Sep 26, 2025 at 9:26 AM Jacob Champion
<[email protected]> wrote:
If it's truly just a few lines, then I've misunderstood what you're
suggesting (patches welcome). It's not the separator splitting that
I'm worried about, but the restructuring of the test.
Here is a very slapdash attempt at pushing the "chunk size iteration"
part of the tests down into the test executable, using null
separators. I think it's ugly, but maybe not quite as bad as I feared.
Unfortunately it only gives about a 4x speedup on my machine, and I
was hoping for much more. (We should really expect this entire thing
to run in a fraction of a second.)
Let me know if you think the tradeoff is worth it for now; I can
polish it up if so.
On my (Linux) test it went from 3.75s to 0.78s, nearly 80% reduction. I
think the reduction on Windows is likely to be more. So I think this is
worth doing.
cheers
andrew
--
Andrew Dunstan
EDB: https://www.enterprisedb.com
