Re: Bug in gawk when both RS="" and file contains only empty lines

[Aharon Robbins]

Greetings.  Re this:

> From: Michel Jouvin <address@hidden>
> Date: Thu, 19 Jul 2001 15:30:34 +0100
> To: address@hidden
> Subject: Bug in gawk when both RS="" and file contains only empty lines
> Cc: address@hidden
>
> I have a problem with gawk on Linux RH 6.2. The problem exists both 
> with the original version (3.0.4) and 3.1 I installed.
>
> I use gawk with the following parameters :
>
>       gawk 'BEGIN {RS=""} {print "Next record :", $0}' input_file
>
> This command prints all the lines of the input file as records (FS=\n 
> implied. the problem remains the same with an explicit FS=\n). This 
> works perfectly as long as the file is empty or conatins one non empty 
> line.
>
> But if it happens that input_file is made only of empty lines (one or 
> several, it doesn't matter but lines should not contain even a space), 
> gawk never exits.
>
> I don't reproduce this behaviour with awk (non Gnu) on other platforms 
> (like Tru64 awk for example) where awk just exits as if the file was 
> empty.
>
> Thanks in advance for any help.
>
> Michel

It's a bug.  Here is an unofficial patch.

Thanks for the bug report.

Arnold

*** ../gawk-3.1.0/io.c  Tue Apr 24 07:35:35 2001
--- io.c        Mon Jul 23 20:24:47 2001
***************
*** 2286,2291 ****
--- 2286,2297 ----
                                 */
                                while (*start == '\n' && start < iop->end)
                                        start++;
+                               /* if file is nothing but newlines, no record 
found */
+                               if (iop->cnt == EOF && start >= iop->end) {
+                                       *out = NULL;
+                                       set_RT_to_null();
+                                       return EOF;
+                               }
                                goto again;
                        }
                        bp = start + RESTART(rsre, start);