lm-verify, long copyright line

lm-verify, long copyright line

[Mike Kupfer]

(lm-verify "mh-e.el") in master complains that the copyright holder is
not the Free Software Foundation, apparently because the copyright
statement wraps:

;; Copyright (C) 1985-1988, 1990, 1992-1995, 1997, 1999-2017 Free
;; Software Foundation, Inc.

https://www.gnu.org/prep/maintain/html_node/Copyright-Notices.html says
that it's okay to split the notice across multiple lines, but lm-verify
apparently doesn't understand that.  Should I ignore the complaint, or
reorganize the statement into multiple lines which each contain the
string "Free Software Foundation", or...?

thanks,
mike

Re: lm-verify, long copyright line

[Paul Eggert]

On 06/22/2017 03:51 PM, Mike Kupfer wrote:
> Should I ignore the complaint, or
> reorganize the statement into multiple lines which each contain the
> string "Free Software Foundation", or...?

I wouldn't reorganize the statement, as it's likely to be re-reorganized 
in January 2018 when we update the copyright years automatically. 
Instead, I'd either ignore the complaint or better yet fix lm-verify 
(the latter should be pretty simple, just change a string match to a 
regex match).

Re: lm-verify, long copyright line

[Mike Kupfer]

Paul Eggert wrote:

> Instead, I'd either ignore the complaint or better yet
> fix lm-verify (the latter should be pretty simple, just change a
> string match to a regex match).

I started looking at lm-verify, and I realized that it expects the
additional line(s) to be indented.  So it's happy with

;; Copyright (C) 1985-1988, 1990, 1992-1995, 1997, 1999-2017 Free
;;   Software Foundation, Inc.

Will the copyright update in 2018 break this?

mike

Re: lm-verify, long copyright line

[Paul Eggert]

On 06/22/2017 04:57 PM, Mike Kupfer wrote:
> Will the copyright update in 2018 break this?

Although I don't know offhand, you should be able to find out by running 
the shell command:

UPDATE_COPYRIGHT_YEAR=2018 admin/update-copyright

on a scratch copy of the source code.

Re: lm-verify, long copyright line

[Mike Kupfer]

Paul Eggert wrote:

> On 06/22/2017 04:57 PM, Mike Kupfer wrote:
> > Will the copyright update in 2018 break this?
> 
> Although I don't know offhand, you should be able to find out by
> running the shell command:
> 
> UPDATE_COPYRIGHT_YEAR=2018 admin/update-copyright
> 
> on a scratch copy of the source code.

The good news is that it successfully changes the final year in mh-e.el
from 2017 to 2018.  The bad news is that it removes the indentation in
the second line.

It looks fairly straightforward to change the update-copyright script to
add some indentation when there's a line wrap.  Would that be
acceptable?

mike

Re: lm-verify, long copyright line

[Paul Eggert]

Mike Kupfer wrote:
> It looks fairly straightforward to change the update-copyright script to
> add some indentation when there's a line wrap.  Would that be
> acceptable?

Quite possibly not, as the script is imported from gnulib and I expect other 
users are used to the current behavior and will see glitches if it starts 
indenting differently. Better would be to add an option to the script to enable 
the behavior you prefer.

However, why does lm-verify insist on unequal indentation in the copyright 
notice? It seems like a weird constraint. Wouldn't it be better to fix that than 
to modify update-copyright? Not every copyright notice is processed by 
update-copyright.

Re: lm-verify, long copyright line

[Stefan Monnier]

> However, why does lm-verify insist on unequal indentation in the copyright
> notice? It seems like a weird constraint.

Because all the lm-* thingies use leading spaces to distinguish line
continuations (along the same lines as what RFC822 does).
It doesn't look weird to me at all.


        Stefan

Re: lm-verify, long copyright line

[Paul Eggert]

On 06/23/2017 09:19 AM, Stefan Monnier wrote:
> Because all the lm-* thingies use leading spaces to distinguish line
> continuations (along the same lines as what RFC822 does).
> It doesn't look weird to me at all.

It's readable either way; all I was saying is that it's weird for 
lm-verify to insist on extra indentation for the second line of a 
copyright notice. The extra indentation isn't that common a style. For 
example, src/emacs.c doesn't use it, nor does src/lisp.h, or 
lisp/subr.el, or lots of other Emacs source files. So why insist on the 
style for lisp/mh-e/mh-e.el?

Re: lm-verify, long copyright line

[Glenn Morris]

Paul Eggert wrote:

> The extra indentation isn't that common a style. For example,
> src/emacs.c doesn't use it, nor does src/lisp.h, or lisp/subr.el, or
> lots of other Emacs source files. 

Emacs source files used it in the past, but the nicer formatting was
lost when gnulib's update-copyright began to be used.
Presumably no-one has run lm-verify since then.

I quibbled at the time! :)

http://lists.gnu.org/archive/html/emacs-devel/2013-12/msg00887.html

Re: lm-verify, long copyright line

[Paul Eggert]

On 06/23/2017 10:59 AM, Glenn Morris wrote:
> Emacs source files used it in the past, but the nicer formatting was
> lost when gnulib's update-copyright began to be used.
> Presumably no-one has run lm-verify since then.

OK, thanks for explaining. The 2013 discussion didn't mention lm-verify 
and so I didn't worry about the indenting change back then. It is an 
annoyance, and I suppose someone should change either lm-verify or 
Gnulib to fix the incompatibility between the two approaches.

Re: lm-verify, long copyright line

[Mike Kupfer]

Paul Eggert wrote:

> I suppose someone should change either
> lm-verify or Gnulib to fix the incompatibility between the two
> approaches.

The change to Gnulib is a one-liner:

--- update-copyright~	2017-06-22 17:16:28.604027480 -0700
+++ update-copyright	2017-06-24 15:08:43.626795139 -0700
@@ -243,7 +243,7 @@
                 || ($stmt =~ s/^([\S]+)(?: |$)//))
               {
                 my $line = $1;
-                $stmt_wrapped .= $stmt_wrapped ? "$eol$prefix" : $leading;
+                $stmt_wrapped .= $stmt_wrapped ? "$eol$prefix  " : $leading;
                 $stmt_wrapped .= $line;
               }
             else

I guess the question is how likely would this be to introduce a new set
of problems.

mike

Re: lm-verify, long copyright line

[Paul Eggert]

[-- Attachment #1: Type: text/plain, Size: 467 bytes --]

Mike Kupfer wrote:
> I guess the question is how likely would this be to introduce a new set
> of problems.

Reasonably likely, I'd say, as the tradition outside Emacs is to indent 
copyright notices the same as other text. Let's make lm-verify a bit more 
generous about copyright notice format; this is a good thing to do regardless of 
whether we change update-copyright. I installed the attached patch into master; 
does it suffice to solve the problem?

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: 0001-Adjust-lm-verify-to-accept-current-notices.patch --]
[-- Type: text/x-patch; name="0001-Adjust-lm-verify-to-accept-current-notices.patch", Size: 1492 bytes --]

From f587a81c5247e79f75c7ec03f4e97a7dd883167b Mon Sep 17 00:00:00 2001
From: Paul Eggert <eggert@cs.ucla.edu>
Date: Sat, 24 Jun 2017 17:54:21 -0700
Subject: [PATCH] Adjust lm-verify to accept current notices

Problem reported by Mike Kupfer in:
http://lists.gnu.org/archive/html/emacs-devel/2017-06/msg00512.html
* lisp/emacs-lisp/lisp-mnt.el (lm-crack-copyright):
Do not require later lines in a copyright notice to have more
indentation than earlier lines.
---
 lisp/emacs-lisp/lisp-mnt.el | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/lisp/emacs-lisp/lisp-mnt.el b/lisp/emacs-lisp/lisp-mnt.el
index fc3caf3..a1c5b69 100644
--- a/lisp/emacs-lisp/lisp-mnt.el
+++ b/lisp/emacs-lisp/lisp-mnt.el
@@ -326,12 +326,13 @@ lm-crack-copyright
 	  (start (point))
 	  (end (line-end-position)))
       ;; Cope with multi-line copyright `lines'.  Assume the second
-      ;; line is indented (with the same commenting style).
+      ;; line is indented at least as much as the original, with the
+      ;; same commenting style.
       (save-excursion
 	(beginning-of-line 2)
-	(let ((str (concat (match-string-no-properties 1) "[ \t]+")))
+	(let ((str (match-string-no-properties 1)))
 	  (beginning-of-line)
-	  (while (looking-at str)
+	  (while (and (looking-at str) (not (looking-at lm-copyright-prefix)))
 	    (setq end (line-end-position))
 	    (beginning-of-line 2))))
       ;; Make a single line and parse that.
-- 
2.7.4

Re: lm-verify, long copyright line

[Mike Kupfer]

Paul Eggert wrote:

> I installed the attached patch into master; does it suffice to solve
> the problem?

Yes, lm-verify no longer complains about

;; Copyright (C) 1985-1988, 1990, 1992-1995, 1997, 1999-2017 Free
;; Software Foundation, Inc.

Thanks!

mike

Re: lm-verify, long copyright line

[Stefan Monnier]

> Reasonably likely, I'd say, as the tradition outside Emacs is to indent
> copyright notices the same as other text. Let's make lm-verify a bit more
> generous about copyright notice format; this is a good thing to do
> regardless of whether we change update-copyright. I installed the attached
> patch into master; does it suffice to solve the problem?

I think the patch is OK.
But I'd still much prefer the continuation lines to be indented.


        Stefan

Re: lm-verify, long copyright line

[Mike Kupfer]

Stefan Monnier wrote:

> But I'd still much prefer the continuation lines to be indented.

This could be achieved by changing the Gnulib update-copyright (the one
in build-aux) to optionally indent.  Control could be via a command-line
flag or an environment variable, which admin/update-copyright could
set.

mike

Re: lm-verify, long copyright line

[Paul Eggert]

Stefan Monnier wrote:
> I'd still much prefer the continuation lines to be indented.

I looked into do that, and it's a bit of a pain. Mike's patch is not quite 
right, since it causes update-copyright to generate lines like this:

-   Copyright (C) 2002, 2004-2007, 2009-2017 Free Software Foundation,
-     Inc.

instead of like this:

+   Copyright (C) 2002, 2004-2007, 2009-2017 Free Software Foundation, Inc.

I guess because update-copyright mistakenly counts the continuation indent as 
being part of the previous line when deciding whether to wrap. No doubt this 
could be fixed, but there's a worse problem: the patch affects all files, even 
those shared with Gnulib where it's better to avoid spurious changes. I suppose 
we could start maintaining a list of the dozens of files imported from Gnulib 
and/or elsewhere (currently we do not have such a list) and having 
update-copyright skip files in that list, but this sounds like quite a bit of 
hassle merely to support a preferred style of indentation for copyright notices.

Re: lm-verify, long copyright line

[Mike Kupfer]

Paul Eggert wrote:

> Stefan Monnier wrote:
> > I'd still much prefer the continuation lines to be indented.
> 
> I looked into do that, and it's a bit of a pain. Mike's patch is not
> quite right, since it causes update-copyright to generate lines like
> this:
> 
> -   Copyright (C) 2002, 2004-2007, 2009-2017 Free Software Foundation,
> -     Inc.
> 
> instead of like this:
> 
> +   Copyright (C) 2002, 2004-2007, 2009-2017 Free Software Foundation, Inc.

I don't understand why the second form is correct, given that
update-copyright has this comment:

#   1. Fit within 72 columns.

mike

Re: lm-verify, long copyright line

[Paul Eggert]

Mike Kupfer wrote:
> I don't understand why the second form is correct, given that
> update-copyright has this comment:
> 
> #   1. Fit within 72 columns.

Ah, sorry, I misinterpreted the output of my tests. You're right, this was not 
an issue with the change. The main problem is the style clash between Emacs's 
own files and files imported from elsewhere.