Re: pgsql: Fix pattern matching logic for logs in TAP tests of pgbench

On 6/24/21 10:12 PM, Andrew Dunstan wrote:
> On 6/24/21 9:53 PM, Michael Paquier wrote:
>> On Thu, Jun 24, 2021 at 09:36:53PM -0400, Andrew Dunstan wrote:
>>> That's not really an equivalent test. I'm taking a look
>> Thanks!
>
>
>
> There's a whole lot wrong with this code. To start with, why is that
> unchecked eval there. And why is it reading in log files on its own
> instead of using TestLib::slurp_file, which, among other things,
> normalizes line endings? There's a very good chance that this latter is
> the issue. It only affects msys which is why you didn't see an issue on
> MSVC. And also, why does it carefully unlink the log files so that any
> trace of what's gone wrong is deleted?
>
>
> Based on the little I've seen this file needs a serious code review.

... and there's the error:

check_pgbench_logs($bdir, '001_pgbench_log_2', 1, 8, 92,
    qr{^[01] \d{1,2} \d+ \d \d+ \d+$});

and one further down the same.

Since the file isn't read in using slurp_file, that $ won't match
because the lines will end \r\n instead of \n.

cheers

andrew

--
Andrew Dunstan
EDB: https://www.enterprisedb.com