Обсуждение: Small fixes for incorrect error messages

Hi all,

I found a few easily overlooked error messages worth fixing, as follows:

01
```
     if (!is_absolute_path(tsmap->new_dir))
-        pg_fatal("old directory is not an absolute path in tablespace mapping: %s",
+        pg_fatal("new directory is not an absolute path in tablespace mapping: %s",
                  tsmap->new_dir);
```

02
```
     if (rb < 0)
-        pg_fatal("could not read from file \"%s\": %m", dst);
+        pg_fatal("could not read from file \"%s\": %m", src);
```

03
```
     if (fstat(fd, &statbuf) < 0)
-        pg_fatal("could not open file \"%s\" for reading: %m",
+        pg_fatal("could not stat file \"%s\" for reading: %m",
                  fullpath);
```

--
regards,
Man Zeng

On Thu, Feb 05, 2026 at 02:51:05PM +0800, zengman wrote:
> I found a few easily overlooked error messages worth fixing, as follows:

Good catches.  Obviously a copy-pasto for the first and third ones,
and a thinko for the second one.
--
Michael

On Thu, Feb 05, 2026 at 07:28:00PM +0900, Michael Paquier wrote:
> Good catches.  Obviously a copy-pasto for the first and third ones,
> and a thinko for the second one.

Fixed these ones.
--
Michael

> Fixed these ones.

Thanks a lot for your help with this! I previously reported another similar issue – could you please take a moment to
lookat it? Maybe we could fix it along the way.
 

https://www.postgresql.org/message-id/flat/tencent_6D253E6F517F8F85796F0D9D@qq.com

--
regards,
Man Zeng

On 05.02.26 07:51, zengman wrote:
> ```
>       if (fstat(fd, &statbuf) < 0)
> -        pg_fatal("could not open file \"%s\" for reading: %m",
> +        pg_fatal("could not stat file \"%s\" for reading: %m",
>                    fullpath);
> ```

It doesn't make sense to "stat a file for reading".  The changed message 
should probably just be

 > +        pg_fatal("could not stat file \"%s\": %m",
 >                    fullpath);

>  It doesn't make sense to "stat a file for reading".  The changed message
>  should probably just be
> +        pg_fatal("could not stat file \"%s\": %m",
>                    fullpath);

You're right. I was careless earlier, my apologies.

--
regards,
Man Zeng

On 07.02.26 10:24, zengman wrote:
>>   It doesn't make sense to "stat a file for reading".  The changed message
>>   should probably just be
>> +        pg_fatal("could not stat file \"%s\": %m",
>>                     fullpath);
> 
> You're right. I was careless earlier, my apologies.

Ok, I have fixed this.

On Sat, Feb 07, 2026 at 11:03:36PM +0100, Peter Eisentraut wrote:
> On 07.02.26 10:24, zengman wrote:
>> You're right. I was careless earlier, my apologies.
>
> Ok, I have fixed this.

Thanks for the fix.  (I was away from my laptop.)
--
Michael