Re: Refactor StringInfo usage in subscriptioncmds.c

2025-11-06 Thread David Rowley
On Fri, 7 Nov 2025 at 01:47, Mats Kindahl wrote: > > I agree with Amit that there doesn't seem to be a need to free > > pubnames.data. We're already leaking publist, for instance. This is > > okay since we only call these functions during DDL, which in general is > > not sensitive to leaks. > >

Re: Refactor StringInfo usage in subscriptioncmds.c

2025-11-06 Thread Mats Kindahl
e. Here is an updated version that removes the pfree() calls. Best wishes, Mats Kindahl From ebf89a82ba463e019e8dba840ddb01905a32b3e7 Mon Sep 17 00:00:00 2001 From: Mats Kindahl Date: Thu, 6 Nov 2025 11:47:50 +0100 Subject: [PATCH] Refactor StringInfo usage in subscriptioncmds.c This commit re

Re: Refactor StringInfo usage in subscriptioncmds.c

2025-11-06 Thread Álvaro Herrera
On 2025-Nov-06, Mats Kindahl wrote: > Attached is a proposal that instead of building error message and hints > dynamically, it will use ereport() directly and as a result does not have to > allocate the error message strings and error message hints dynamically and > these can be removed. LGTM, t

Re: Refactor StringInfo usage in subscriptioncmds.c

2025-11-06 Thread Mats Kindahl
On 11/6/25 13:09, Amit Kapila wrote: On Thu, Nov 6, 2025 at 5:24 PM David Rowley wrote: On Fri, 7 Nov 2025 at 00:38, Mats Kindahl wrote: As discussed in [1] the functions check_publications_origin_tables() and check_publications_origin_sequences() are building error messages using dynamically

Re: Refactor StringInfo usage in subscriptioncmds.c

2025-11-06 Thread Amit Kapila
On Thu, Nov 6, 2025 at 5:24 PM David Rowley wrote: > > On Fri, 7 Nov 2025 at 00:38, Mats Kindahl wrote: > > As discussed in [1] the functions check_publications_origin_tables() and > > check_publications_origin_sequences() are building error messages using > > dynamically allocated StringInfo ins

Re: Refactor StringInfo usage in subscriptioncmds.c

2025-11-06 Thread David Rowley
On Fri, 7 Nov 2025 at 00:38, Mats Kindahl wrote: > As discussed in [1] the functions check_publications_origin_tables() and > check_publications_origin_sequences() are building error messages using > dynamically allocated StringInfo instances only to avoid duplicating a > call of ereport(). Looks

Refactor StringInfo usage in subscriptioncmds.c

2025-11-06 Thread Mats Kindahl
57edc From 2436de272d39e664fa0c0010cf875cb34fcecd1e Mon Sep 17 00:00:00 2001 From: Mats Kindahl Date: Thu, 6 Nov 2025 11:47:50 +0100 Subject: [PATCH] Refactor StringInfo usage in subscriptioncmds.c This commit removes some uses of StringInfo that was not necessary and was only there to avoid having to write two calls