Hi,

> I am afraid that would generate a lot of noises for future reviews.
>
> 2. A typo in the patch
>
> ```
> +    # Check each line except for the fist and the last one
> ```

Thanks, fixed.

> 3. As you are updating pgindent, I want to report an issue, you may address 
> in a separate patch or just in this patch, up to you.
>
> See this code:
> ```
>             else
>                  /*
>                   * fetch all the rest of the page
>                   */
>                  copysize = QUEUE_PAGESIZE - curoffset;
> ```
>
> In the “else” clause, there is a multiple-line comment block, and a single 
> line of code. Pgindent will add an empty line between “else” and the comment 
> block, which is weird. If the comment is one-line, then no empty line will be 
> inserted.

I didn't manage to find this code. The closest thing I see is in
src/backend/commands/async.c:

```
            else
            {
                /* fetch all the rest of the page */
                copysize = QUEUE_PAGESIZE - curoffset;
            }
```

Last time it was touched 15 years ago and pgindent processes it as expected.

> 1. I just ran the patched pgindent against a random file, then I got a lot 
> diffs like:
>
> ```
>         /*
> -        * Direct advancement: avoid waking non-caught up backends that
> -        * aren't interested in our notifications.
> +        * Direct advancement: avoid waking non-caught up backends that aren't
> +        * interested in our notifications.
>          */
> ```

I'm not sure if this is part of the PostgreSQL code base either. My
best guess is that something is wrong with whitespaces here (tabs vs
spaces). We have plenty of multiline comments like this and from what
I can tell they are processed correctly.

-- 
Best regards,
Aleksander Alekseev
From ad91fda8a7406f9fd5dbbfe5a895a811230e2eec Mon Sep 17 00:00:00 2001
From: Aleksander Alekseev <[email protected]>
Date: Fri, 20 Jun 2025 16:31:36 +0300
Subject: [PATCH v6] pgindent: improve formatting of multiline comments

Format multiline comments like this:

/* line 1
 * line 2
 */

... into:

/*
 * line 1
 * line 2
 */

This is more consistent with what we currently have in the tree.

Author: Aleksander Alekseev <[email protected]>
Reported-by: Michael Paquier <[email protected]>
Reviewed-by: Arseniy Mukhin <[email protected]>
Reviewed-by: Nathan Bossart <[email protected]>
Discussion: https://postgr.es/m/CAJ7c6TPQ0kkHQG-AqeAJ3PV_YtmDzcc7s%2B_V4%3Dt%2BxgSnZm1cFw%40mail.gmail.com
---
 src/tools/pgindent/pgindent | 36 ++++++++++++++++++++++++++++++++++++
 1 file changed, 36 insertions(+)

diff --git a/src/tools/pgindent/pgindent b/src/tools/pgindent/pgindent
index b7d71808924..4db12cb1d92 100755
--- a/src/tools/pgindent/pgindent
+++ b/src/tools/pgindent/pgindent
@@ -281,6 +281,9 @@ sub post_indent
 	# Fix run-together comments to have a tab between them
 	$source =~ s!\*/(/\*.*\*/)$!*/\t$1!gm;
 
+	# Postprocess multiline comments except for /**... and /*-... ones
+	$source =~ s!^(/\*[^\*\-].*?\*/)!postprocess_multiline_comment($1)!mgse;
+
 	## Functions
 
 	# Use a single space before '*' in function return types
@@ -289,6 +292,39 @@ sub post_indent
 	return $source;
 }
 
+sub postprocess_multiline_comment
+{
+    my $source = shift;
+    my @lines = split "\n", $source;
+
+    # Only format comments that match the expected format,
+    # or at least that could have been the author's intent.
+    if (($lines[0] ne "/*" && $lines[-1] ne " */") or ($lines[1] !~ m!^ \*!))
+    {
+        return $source;
+    }
+
+    # Check each line except for the first and the last one
+    for my $i ( 1 .. scalar @lines - 2 )
+    {
+        $lines[$i] = " *".$lines[$i] if $lines[$i] !~ /^ \*/;
+    }
+
+    # Keep /* === and /* --- lines as is
+    if ($lines[0] !~ m!^/\* [=-]+!) {
+        $lines[0] =~ s!/\*(.+)!/\*\n *$1!;
+    }
+
+    # Keep === */ and --- */ lines as is
+    if ($lines[-1] !~ m![=-]+ \*/$!) {
+        $lines[-1] =~ s!(.+) \*/!$1\n \*/!;
+    }
+
+    $source = join "\n", @lines;
+
+    return $source;
+}
+
 sub run_indent
 {
 	my $source = shift;
-- 
2.43.0

Reply via email to