Re: [Windows,PATCH] Use faster, higher precision timer API - Mailing list pgsql-hackers

From Tom Lane
Subject Re: [Windows,PATCH] Use faster, higher precision timer API
Date
Msg-id 5004.1410971939@sss.pgh.pa.us
Whole thread Raw
In response to Re: [Windows,PATCH] Use faster, higher precision timer API  (Andrew Dunstan <andrew@dunslane.net>)
Responses Re: [Windows,PATCH] Use faster, higher precision timer API  (Andres Freund <andres@2ndquadrant.com>)
List pgsql-hackers
Andrew Dunstan <andrew@dunslane.net> writes:
> On 09/17/2014 08:27 AM, Craig Ringer wrote:
>> Attached is a patch to switch 9.5 over to using the
>> GetSystemTimeAsFileTime call instead of separate GetSystemTime and
>> SystemTimeToFileTime calls.

> That will presumably breaK XP. I know XP has been declared at EOL, but 
> there are still a heck of a lot of such systems out there,

Yeah.  Do we really think more precise timestamps are worth dropping
XP support?  On the Unix side, I know exactly what would happen to a
patch proposing that we replace gettimeofday() with clock_gettime()
with no thought for backwards compatibility.  Why would we expect
less on the Windows side?

Quite aside from XP ... AFAICS from the patch description, this patch
in itself moves us to a place that's a net negative in terms of
functionality.  Maybe it's a stepping stone to something better,
but I think we should just go directly to the something better.
I don't care for committing regressions on the promise that they'll
get fixed later.

Or in short: let's do the work needed to adapt our code to what's
available on the particular Windows version *first*.  Once we've
got that configuration support done, it shouldn't be much extra
work to continue XP support here.
        regards, tom lane



pgsql-hackers by date:

Previous
From: Marti Raudsepp
Date:
Subject: Re: Join consolidation / Removing duplicate joins
Next
From: Andres Freund
Date:
Subject: Re: [Windows,PATCH] Use faster, higher precision timer API