Re: Ryu floating point output patch

Andrew Dunstan <andrew.dunstan@2ndquadrant.com>

From: Andrew Dunstan <andrew.dunstan@2ndquadrant.com>
To: Andrew Gierth <andrew@tao11.riddles.org.uk>, Tom Lane <tgl@sss.pgh.pa.us>
Cc: pgsql-hackers@postgresql.org, "David G. Johnston" <david.g.johnston@gmail.com>, Donald Dong <xdong@csumb.edu>, Andres Freund <andres@anarazel.de>, Robert Haas <robertmhaas@gmail.com>, Chapman Flack <chap@anastigmatix.net>
Date: 2019-02-17T16:49:50Z
Lists: pgsql-hackers
On 2/17/19 11:19 AM, Andrew Gierth wrote:
>>>>>> "Tom" == Tom Lane <tgl@sss.pgh.pa.us> writes:
>  >> b) new command-line option, e.g. pg_dump --extra-float-digits=0
>
>  >> This is probably the safest option, IMO. Any preferences as to the
>  >> option name?
>
>  Tom> I like that too, assuming that it can be made to fit into the
>  Tom> structure of the cross-version-upgrade tests (which would have
>  Tom> to know to use it only with >= v12 pg_dump).
>
>  Tom> --extra-float-digits seems fine as a name.
>
> As far as I can tell, test.sh only uses the new version's pg_dump (in
> fact the only binaries directly used from the old version are initdb and
> pg_ctl).
>
> ... wait, this initdb in test.sh is using an option that doesn't exist
> before pg11? Is the cross-version test actually using some different
> script? Andrew?
>



The buildfarm cross-version upgrade tests do not run test.sh. See
<https://github.com/PGBuildFarm/client-code/blob/master/PGBuild/Modules/TestUpgradeXversion.pm>


Essentially it takes the result of a previous buildfarm run on the
source branch, dumps it, upgrades it, and dumps again, then compares the
two dumps. It doesn't itself run the regression tests.


cheers


andrew


-- 
Andrew Dunstan                https://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services



Commits

  1. Provide an extra-float-digits setting for pg_dump / pg_dumpall

  2. Change floating-point output format for improved performance.

  3. Use strtof() and not strtod() for float4 input.