agora inbox for pgsql-hackers@postgresql.org  
help / color / mirror / Atom feed
From: Nathan Bossart <nathandbossart@gmail.com>
To: Chao Li <li.evan.chao@gmail.com>
Cc: Tom Lane <tgl@sss.pgh.pa.us>
Cc: pgsql-hackers@postgresql.org
Subject: Re: postgresql.conf.sample tab width
Date: Sun, 30 Aug 2026 11:48:12 -0500
Message-ID: <apRezMywHRuR5h-6@nathan> (raw)
In-Reply-To: <aRyf9XOgRhHn-AiO@nathan>
References: <aReNUKdMgKxLqmq7@nathan>
	<3230808.1763151915@sss.pgh.pa.us>
	<aRefX8JOs_3VbErE@nathan>
	<3238135.1763156048@sss.pgh.pa.us>
	<aRehOVyEMFIxl5rE@nathan>
	<aRtugij0CQTbpmIx@nathan>
	<EE0918D4-F05C-4047-910C-F0DED136ADE7@gmail.com>
	<aRyf9XOgRhHn-AiO@nathan>

I noticed that initdb still adds tabs to postgresql.conf.  Here is a patch
to fix that, which simplifies replace_guc_value() a bit.

-- 
nathan
From 7894bcefd428bd77c926a605b620e946a5b86ade Mon Sep 17 00:00:00 2001
From: Nathan Bossart <nathan@postgresql.org>
Date: Sun, 30 Aug 2026 11:41:41 -0500
Subject: [PATCH v1 1/1] initdb: Pad rewritten GUC lines with spaces, not tabs.

Commit f63ae72bbc converted postgresql.conf.sample to spaces so
that it lines up at any tab width, but replace_guc_value() still
pads with tabs when it re-aligns the trailing comment of a setting
it rewrites.  The postgresql.conf that initdb generates therefore
mixes the two.

To fix, pad with spaces.  While at it, simplify the calculation of
the original comment column: with no tabs left in the sample file
or in the lines we write out in its place, the de-tab-ifying loop
is only ever counting bytes, so have it count bytes directly.

Oversight in commit f63ae72bbc.

Discussion: https://postgr.es/m/aReNUKdMgKxLqmq7%40nathan
Backpatch-through: 19
---
 src/bin/initdb/initdb.c | 31 +++++--------------------------
 1 file changed, 5 insertions(+), 26 deletions(-)

diff --git a/src/bin/initdb/initdb.c b/src/bin/initdb/initdb.c
index b3d496372ad..86d42a98a27 100644
--- a/src/bin/initdb/initdb.c
+++ b/src/bin/initdb/initdb.c
@@ -587,37 +587,16 @@ replace_guc_value(char **lines, const char *guc_name, const char *guc_value,
 		{
 			/*
 			 * We try to preserve original indentation, which is tedious.
-			 * oldindent and newindent are measured in de-tab-ified columns.
 			 */
-			const char *ptr;
-			int			oldindent = 0;
-			int			newindent;
+			int			oldindent = (int) (where - lines[i]);
+			int			newindent = newline->len;
 
-			for (ptr = lines[i]; ptr < where; ptr++)
-			{
-				if (*ptr == '\t')
-					oldindent += 8 - (oldindent % 8);
-				else
-					oldindent++;
-			}
-			/* ignore the possibility of tabs in guc_value */
-			newindent = newline->len;
-			/* append appropriate tabs and spaces, forcing at least one */
+			/* append appropriate spaces, forcing at least one */
 			oldindent = Max(oldindent, newindent + 1);
 			while (newindent < oldindent)
 			{
-				int			newindent_if_tab = newindent + 8 - (newindent % 8);
-
-				if (newindent_if_tab <= oldindent)
-				{
-					appendPQExpBufferChar(newline, '\t');
-					newindent = newindent_if_tab;
-				}
-				else
-				{
-					appendPQExpBufferChar(newline, ' ');
-					newindent++;
-				}
+				appendPQExpBufferChar(newline, ' ');
+				newindent++;
 			}
 			/* and finally append the old comment */
 			appendPQExpBufferStr(newline, where);
-- 
2.55.0

Attachments:

  [text/plain] v1-0001-initdb-Pad-rewritten-GUC-lines-with-spaces-not-ta.patch (2.4K, ../apRezMywHRuR5h-6@nathan/2-v1-0001-initdb-Pad-rewritten-GUC-lines-with-spaces-not-ta.patch)
  download | inline diff:
From 7894bcefd428bd77c926a605b620e946a5b86ade Mon Sep 17 00:00:00 2001
From: Nathan Bossart <nathan@postgresql.org>
Date: Sun, 30 Aug 2026 11:41:41 -0500
Subject: [PATCH v1 1/1] initdb: Pad rewritten GUC lines with spaces, not tabs.

Commit f63ae72bbc converted postgresql.conf.sample to spaces so
that it lines up at any tab width, but replace_guc_value() still
pads with tabs when it re-aligns the trailing comment of a setting
it rewrites.  The postgresql.conf that initdb generates therefore
mixes the two.

To fix, pad with spaces.  While at it, simplify the calculation of
the original comment column: with no tabs left in the sample file
or in the lines we write out in its place, the de-tab-ifying loop
is only ever counting bytes, so have it count bytes directly.

Oversight in commit f63ae72bbc.

Discussion: https://postgr.es/m/aReNUKdMgKxLqmq7%40nathan
Backpatch-through: 19
---
 src/bin/initdb/initdb.c | 31 +++++--------------------------
 1 file changed, 5 insertions(+), 26 deletions(-)

diff --git a/src/bin/initdb/initdb.c b/src/bin/initdb/initdb.c
index b3d496372ad..86d42a98a27 100644
--- a/src/bin/initdb/initdb.c
+++ b/src/bin/initdb/initdb.c
@@ -587,37 +587,16 @@ replace_guc_value(char **lines, const char *guc_name, const char *guc_value,
 		{
 			/*
 			 * We try to preserve original indentation, which is tedious.
-			 * oldindent and newindent are measured in de-tab-ified columns.
 			 */
-			const char *ptr;
-			int			oldindent = 0;
-			int			newindent;
+			int			oldindent = (int) (where - lines[i]);
+			int			newindent = newline->len;
 
-			for (ptr = lines[i]; ptr < where; ptr++)
-			{
-				if (*ptr == '\t')
-					oldindent += 8 - (oldindent % 8);
-				else
-					oldindent++;
-			}
-			/* ignore the possibility of tabs in guc_value */
-			newindent = newline->len;
-			/* append appropriate tabs and spaces, forcing at least one */
+			/* append appropriate spaces, forcing at least one */
 			oldindent = Max(oldindent, newindent + 1);
 			while (newindent < oldindent)
 			{
-				int			newindent_if_tab = newindent + 8 - (newindent % 8);
-
-				if (newindent_if_tab <= oldindent)
-				{
-					appendPQExpBufferChar(newline, '\t');
-					newindent = newindent_if_tab;
-				}
-				else
-				{
-					appendPQExpBufferChar(newline, ' ');
-					newindent++;
-				}
+				appendPQExpBufferChar(newline, ' ');
+				newindent++;
 			}
 			/* and finally append the old comment */
 			appendPQExpBufferStr(newline, where);
-- 
2.55.0

view thread (12+ messages)  latest in thread

Message-ID: <apRezMywHRuR5h-6@nathan>
Permalink:  ../apRezMywHRuR5h-6@nathan/
Also on:    postgresql.org/message-id/apRezMywHRuR5h-6@nathan

reply

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Reply to all the recipients using the --to and --cc options:
  reply via email

  To: pgsql-hackers@postgresql.org
  Cc: nathandbossart@gmail.com, li.evan.chao@gmail.com, tgl@sss.pgh.pa.us
  Subject: Re: postgresql.conf.sample tab width
  In-Reply-To: <apRezMywHRuR5h-6@nathan>

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

This inbox is served by agora; see mirroring instructions
for how to clone and mirror all data and code used for this inbox