Received: from malur.postgresql.org ([217.196.149.56]) by arkaria.postgresql.org with esmtps (TLS1.3:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.92) (envelope-from ) id 1o8PwL-0000Lw-UI for pgsql-hackers@arkaria.postgresql.org; Mon, 04 Jul 2022 17:32:53 +0000 Received: from localhost ([127.0.0.1] helo=malur.postgresql.org) by malur.postgresql.org with esmtp (Exim 4.92) (envelope-from ) id 1o8PwJ-0005QH-HA for pgsql-hackers@arkaria.postgresql.org; Mon, 04 Jul 2022 17:32:51 +0000 Received: from magus.postgresql.org ([2a02:c0:301:0:ffff::29]) by malur.postgresql.org with esmtps (TLS1.3:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.92) (envelope-from ) id 1o8PwJ-0005Q1-89 for pgsql-hackers@lists.postgresql.org; Mon, 04 Jul 2022 17:32:51 +0000 Received: from sss.pgh.pa.us ([66.207.139.130]) by magus.postgresql.org with esmtps (TLS1.3:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.92) (envelope-from ) id 1o8PwE-0000Rn-8o for pgsql-hackers@postgresql.org; Mon, 04 Jul 2022 17:32:47 +0000 Received: from sss1.sss.pgh.pa.us (localhost [127.0.0.1]) by sss.pgh.pa.us (8.15.2/8.15.2) with ESMTP id 264HWhjM1598592; Mon, 4 Jul 2022 13:32:43 -0400 From: Tom Lane To: Peter Eisentraut cc: pgsql-hackers Subject: Re: pgsql: Change timeline field of IDENTIFY_SYSTEM to int8 In-reply-to: References: <1467434.1656914113@sss.pgh.pa.us> Comments: In-reply-to Peter Eisentraut message dated "Mon, 04 Jul 2022 15:55:30 +0200" MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-ID: <1598590.1656955963.1@sss.pgh.pa.us> Date: Mon, 04 Jul 2022 13:32:43 -0400 Message-ID: <1598591.1656955963@sss.pgh.pa.us> List-Id: List-Help: List-Subscribe: List-Post: List-Owner: List-Archive: Archived-At: Precedence: bulk Peter Eisentraut writes: > On 04.07.22 07:55, Tom Lane wrote: >> But what about whatever code is reading the output? And what if >> that code isn't v16? I can't believe that we can make a wire >> protocol change as summarily as this. > I think a client will either just read the string value and convert it > to some numeric type without checking what type was actually sent, or if > the client API is type-aware and automatically converts to a native type > of some sort, then it will probably already support 64-bit ints. Do you > see some problem scenario? If the result of IDENTIFY_SYSTEM is always sent in text format, then I agree that this isn't very problematic. If there are any clients that fetch it in binary mode, though, this is absolutely a wire protocol break for them ... and no, I don't believe an unsupported claim that they'd adapt automatically. > I'm seeing a bigger problem now, which is that our client code doesn't > parse bigger-than-int32 timeline IDs correctly. Yup. regards, tom lane