GNU bug report logs - #22816
cmp --verbose EOF message could be more verbose

Previous Next

To add a comment to this bug, you must first unarchive it, by sending
a message to control AT debbugs.gnu.org, with unarchive 22816 in the body.
You can then email your comments to 22816 AT debbugs.gnu.org in the normal way.

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

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

From: 積丹尼 Dan Jacobson <jidanni <at> jidanni.org>
To: bug-diffutils <at> gnu.org
Subject: cmp --verbose EOF message could be more verbose
Date: Fri, 26 Feb 2016 17:53:07 +0800

One would think the "cmp: EOF on SHORTER-FILE" message
should at least say what line number if --verbose or --print-bytes is given.

Message #8 received at 22816 <at> debbugs.gnu.org (full text, mbox):

From: Jim Meyering <jim <at> meyering.net>
To: 積丹尼 Dan Jacobson <jidanni <at> jidanni.org>
Cc: 22816 <at> debbugs.gnu.org
Subject: Re: [bug-diffutils] bug#22816: cmp --verbose EOF message could be
 more verbose
Date: Sat, 30 Apr 2016 19:55:21 -0700

On Fri, Feb 26, 2016 at 1:53 AM, 積丹尼 Dan Jacobson <jidanni <at> jidanni.org> wrote:
> One would think the "cmp: EOF on SHORTER-FILE" message
> should at least say what line number if --verbose or --print-bytes is given.

Good idea.
Patch welcome, but it need not hold up a release.

Message #13 received at 22816-done <at> debbugs.gnu.org (full text, mbox):

From: Jim Meyering <jim <at> meyering.net>
To: 22816-done <at> debbugs.gnu.org, Dan Jacobson <jidanni <at> jidanni.org>
Subject: Re: cmp --verbose EOF message could be more verbose
Date: Thu, 4 May 2017 10:28:51 -0700

tags 22816 + wontfix
done

Re https://debbugs.gnu.org/22816, I realized that printing any line
number upon EOF would necessitate counting them from the start, and I
would rather not do that.

If you encounter EOF in this case and want to know how many newlines
were in one of those inputs, you can run "wc -l" on that input.

Making cmp perform that task does not seem worthwhile, because one
would have to choose between always counting newlines (like wc -l
does), and thus incurring this added cost all the time, or accepting a
partial (and racy) solution of rereading the regular file upon EOF. I
say "partial" because if the input is not seekable, that is not even
possible. And "racy" because once rewound and reread, we may then be
processing different content.

So I'm closing this as WONTFIX.

Message #18 received at 22816 <at> debbugs.gnu.org (full text, mbox):

From: Paul Eggert <eggert <at> cs.ucla.edu>
To: jim <at> meyering.net
Cc: 22816 <at> debbugs.gnu.org, jidanni <at> jidanni.org
Subject: Re: [bug-diffutils] bug#22816: cmp --verbose EOF message could be
 more verbose
Date: Thu, 4 May 2017 17:24:19 -0700

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

On 05/04/2017 10:28 AM, Jim Meyering wrote:
> Making cmp perform that task does not seem worthwhile, because one
> would have to choose between always counting newlines (like wc -l
> does), and thus incurring this added cost all the time, or accepting a
> partial (and racy) solution of rereading the regular file upon EOF.

There's a third way, that satisfies Dan Jacobson's request in the usual 
case, without incurring added cost. Proposed patch attached, with a 
simple test for 'cmp' (which is a good thing to have in any event).

[0001-cmp-report-prefix-length-when-one-file-is-shorter.patch (text/x-patch, attachment)]

Message #21 received at 22816 <at> debbugs.gnu.org (full text, mbox):

From: Jim Meyering <jim <at> meyering.net>
To: Paul Eggert <eggert <at> cs.ucla.edu>
Cc: 22816 <at> debbugs.gnu.org, Dan Jacobson <jidanni <at> jidanni.org>
Subject: Re: [bug-diffutils] bug#22816: cmp --verbose EOF message could be
 more verbose
Date: Thu, 4 May 2017 18:14:17 -0700

On Thu, May 4, 2017 at 5:24 PM, Paul Eggert <eggert <at> cs.ucla.edu> wrote:
> On 05/04/2017 10:28 AM, Jim Meyering wrote:
>>
>> Making cmp perform that task does not seem worthwhile, because one
>> would have to choose between always counting newlines (like wc -l
>> does), and thus incurring this added cost all the time, or accepting a
>> partial (and racy) solution of rereading the regular file upon EOF.
>
>
> There's a third way, that satisfies Dan Jacobson's request in the usual
> case, without incurring added cost. Proposed patch attached, with a simple
> test for 'cmp' (which is a good thing to have in any event).

Oh! Glad you did that. I (now) see that cmp already did compute line numbers.
Thanks a lot for adding the tests, too.

My only question is whether to print zero-indexed line and byte
numbers. I think most people will interpret "after byte 5, line 1" as
the 5th byte on the first line, but with that patch, it means the 6th
byte on the second line.
I.e., what should this print?

  cmp <(printf '1\nfoo') <(printf '1\nfoolery\n')

Message #24 received at 22816 <at> debbugs.gnu.org (full text, mbox):

From: Paul Eggert <eggert <at> cs.ucla.edu>
To: Jim Meyering <jim <at> meyering.net>
Cc: 22816 <at> debbugs.gnu.org, Dan Jacobson <jidanni <at> jidanni.org>
Subject: Re: [bug-diffutils] bug#22816: cmp --verbose EOF message could be
 more verbose
Date: Fri, 5 May 2017 08:01:16 -0700

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

Jim Meyering wrote:
> what should this print?
>
>   cmp <(printf '1\nfoo') <(printf '1\nfoolery\n')

Good point, I didn't think of that.  Also, now that I think about it more, a 
diagnostic ending in "after byte 0, line 0" is confusing in a command that 
ordinarily uses origin-1 offsets.

I installed the attached further patch.  With it, your example now outputs:

cmp: EOF on /dev/fd/63 after byte 5, line 2

This is more consistent with longstanding behavior:

$ cmp <(printf '1\nfoot') <(printf '1\nfoolery\n')
/dev/fd/63 /dev/fd/62 differ: byte 6, line 2

[0001-cmp-improve-EOF-diagnostic.patch (text/x-diff, attachment)]

Previous Next

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