GNU bug report logs - #49590
Error in GNU sed manual

Previous Next

To reply to this bug, email your comments to 49590 AT debbugs.gnu.org.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Matthew Hoffman <matthewchoffman <at> protonmail.com>
To: bug-sed <at> gnu.org
Subject: Error in GNU sed manual
Date: Fri, 16 Jul 2021 00:45:07 +0000

[Message part 1 (text/plain, inline)]

Dear GNU people:

The GNU sed manual seems to have an error in it.

In this section:

6.3 Multiline techniques - using D,G,H,N,P to
process multiple lines

You have this example code:

sed '/./{H;$!d} ; x ; s/^/\nSTART-->/ ; s/$/\n<--END/' input.txt

The input file you give is:

a a a aa aaa
aaaa aaaa aa
aaaa aaa aaa
bbbb bbb bbb
bb bb bbb bb
bbbbbbbb bbb
ccc ccc cccc
cccc ccccc c
cc cc cc cc

and the output promised is:

START-->
a a a aa aaa
aaaa aaaa aa
aaaa aaa aaa
<--END
START-->
bbbb bbb bbb
bb bb bbb bb
bbbbbbbb bbb
<--END
START-->
ccc ccc cccc
cccc ccccc c
cc cc cc cc
<--END

This seemed wrong to me, because as I understood it the '$' symbol should indicate the last line of the file, not merely a blank line, so I executed it on my GNU sed 4.7, and this is the output you get:

START-->
a a a aa aaa
aaaa aaaa aa
aaaa aaa aaa
bbbb bbb bbb
bb bb bbb bb
bbbbbbbb bbb
ccc ccc cccc
cccc ccccc c
cc cc cc cc
<--END

This is what I expected based on other documentation, and it is what happened.

I thought you would want to know.

God bless,
Matthew Hoffman

[Message part 2 (text/html, inline)]

Message #8 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Davide Brini <dave_br <at> gmx.com>
To: bug-sed <at> gnu.org
Cc: Matthew Hoffman <matthewchoffman <at> protonmail.com>
Subject: Re: bug#49590: Error in GNU sed manual
Date: Fri, 16 Jul 2021 16:12:34 +0200

On Fri, 16 Jul 2021 00:45:07 +0000, Matthew Hoffman via <bug-sed <at> gnu.org>
wrote:

> Dear GNU people:
>
> The GNU sed manual seems to have an error in it.
>
> In this section:
>
> 6.3 Multiline techniques - using D,G,H,N,P to
> process multiple lines
>
> You have this example code:
>
> sed '/./{H;$!d} ; x ; s/^/\nSTART-->/ ; s/$/\n<--END/' input.txt
>
> The input file you give is:
>
> a a a aa aaa
> aaaa aaaa aa
> aaaa aaa aaa
> bbbb bbb bbb
> bb bb bbb bb
> bbbbbbbb bbb
> ccc ccc cccc
> cccc ccccc c
> cc cc cc cc

Not really. The input file is:

--- start here -----
a a a aa aaa
aaaa aaaa aa
aaaa aaa aaa

bbbb bbb bbb
bb bb bbb bb
bbbbbbbb bbb

ccc ccc cccc
cccc ccccc c
cc cc cc cc

--- end here ----

empty lines are important.

> and the output promised is:
>
> START-->
> a a a aa aaa
> aaaa aaaa aa
> aaaa aaa aaa
> <--END
> START-->
> bbbb bbb bbb
> bb bb bbb bb
> bbbbbbbb bbb
> <--END
> START-->
> ccc ccc cccc
> cccc ccccc c
> cc cc cc cc
> <--END

Again, you missed the empty lines.

> This seemed wrong to me, because as I understood it the '$' symbol should
> indicate the last line of the file, not merely a blank line, so I
> executed it on my GNU sed 4.7, and this is the output you get:
>
> START-->
> a a a aa aaa
> aaaa aaaa aa
> aaaa aaa aaa
> bbbb bbb bbb
> bb bb bbb bb
> bbbbbbbb bbb
> ccc ccc cccc
> cccc ccccc c
> cc cc cc cc
> <--END

With the (incorrect) input file you showed, yes, this is what you get.
With the actual file, you get the output shown in the manual.

BTW, in this case the dollar sign ($) is not representing the last line of
the file, but the end of a buffer. Here's a brief explanation of what the
code does:

/./{H;$!d}
This loads a full "paragraph" (ie lines that come before an empty line)
into the hold buffer. Remember that "d" starts a new cycle.

x
This inverts the pattern and the hold buffer, so the paragraph is in the
pattern buffer.

s/^/\nSTART-->/ ; s/$/\n<--END/
This merely prepends '\nSTART-->' and appends '\n<--END' to a paragraph
(with literal newlines). Here, ^ represent the beginning of the buffer, and
$ the end of the buffer.

Finally, since by default sed prints the contents of the pattern buffer at
the end of a cycle, the decorated paragraph is then printed.

--
D.

Message #11 received at 49590 <at> debbugs.gnu.org (full text, mbox):

From: Jim Meyering <jim <at> meyering.net>
To: Davide Brini <dave_br <at> gmx.com>
Cc: Matthew Hoffman <matthewchoffman <at> protonmail.com>, 49590 <at> debbugs.gnu.org
Subject: Re: bug#49590: Error in GNU sed manual
Date: Sat, 17 Jul 2021 14:15:40 -0700

tags 49590 notabug
close 49590
done

Thanks, Davide, for explaining. I'm closing this and marking it as a non-bug.

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.