Discussion of RFC 2822 ()

Pete Resnick | 26 Sep 2008 00:26

On ABNF in 5322: Advice needed


5322 (nee 2822upd) is in AUTH48. During this process, the editor 
noted some confusion about some of the text: Sometimes when I say 
(e.g.) "phrase", it's not clear whether I'm referring to the ABNF 
terminal "phrase" or a description of it. It's somewhat easy to 
distinguish, for instance, "a quoted string" from "quoted-string", 
but other places it's trickier.

The suggestion was made (and it's been made by others) that all ABNF 
terminals should be surrounded with angle brackets, so it's easy to 
distinguish "a phrase" from "<phrase>" and "a quoted string" from 
"<quoted-string>". My argument against doing this has always been (1) 
there are very few places in the text where it is ambiguous, and none 
(AFAIK) where it makes a difference; and (2) going through and making 
the change could result in errors. (It's a lot of occurrences to go 
through.) Advice sought:

Choice (A): "Don't touch a damn thing, Pete! It's too much of a risk 
to screw things up, and we're moving to Draft Standard. And nothing 
is so ambiguous as to make any difference."

Choice (B): "You absolutely MUST make the change, Pete! The 
ambiguities are serious enough, and there is little risk of making 
things any worse than they are now."

Choice (C): "I've got no horse in this race. Do whatever makes sense 
and that everyone else is up for."

Note for the record that the following is *NOT* an acceptable choice:

(Continue reading)

SM | 26 Sep 2008 16:39

Re: On ABNF in 5322: Advice needed


Hi Pete,
At 15:26 25-09-2008, Pete Resnick wrote:
>5322 (nee 2822upd) is in AUTH48. During this process, the editor 
>noted some confusion about some of the text: Sometimes when I say 
>(e.g.) "phrase", it's not clear whether I'm referring to the ABNF 
>terminal "phrase" or a description of it. It's somewhat easy to 
>distinguish, for instance, "a quoted string" from "quoted-string", 
>but other places it's trickier.
>
>The suggestion was made (and it's been made by others) that all ABNF 
>terminals should be surrounded with angle brackets, so it's easy to 
>distinguish "a phrase" from "<phrase>" and "a quoted string" from 
>"<quoted-string>". My argument against doing this has always been 
>(1) there are very few places in the text where it is ambiguous, and 
>none (AFAIK) where it makes a difference; and (2) going through and 
>making the change could result in errors. (It's a lot of occurrences 
>to go through.) Advice sought:
>
>Choice (A): "Don't touch a damn thing, Pete! It's too much of a risk 
>to screw things up, and we're moving to Draft Standard. And nothing 
>is so ambiguous as to make any difference."

I read the I-D again and I don't the need for a change at this stage 
especially as there is the risk to screw things up.  The editor has a 
different perspective reading the I-D and would find ambiguities we 
might not see.  I'm for choice (A).  I'll rely on your best judgement 
to determine what could be done in the few places where it is ambiguous.

>If you choose (B), I'd also like to hear your answer to the

(Continue reading)

Bill McQuillan | 26 Sep 2008 10:50

Picon

Re: On ABNF in 5322: Advice needed

On Thu, 2008-09-25, Pete Resnick wrote:

> 5322 (nee 2822upd) is in AUTH48. During this process, the editor 
> noted some confusion about some of the text: Sometimes when I say 
> (e.g.) "phrase", it's not clear whether I'm referring to the ABNF 
> terminal "phrase" or a description of it. It's somewhat easy to 
> distinguish, for instance, "a quoted string" from "quoted-string", 
> but other places it's trickier.

Quoting from RFC 5234, "Augmented BNF for Syntax Specifications", 2.1:

   Unlike original BNF, angle brackets ("<", ">") are not required.
   However, angle brackets may be used around a rule name whenever their
   presence facilitates in discerning the use of a rule name.  This is
   typically restricted to rule name references in free-form prose[...]

This seems to me to allow angle brackets to be used in only those places
where it is needed to remove ambiguity.

If I understand you correctly, you have already found a small number of
places where ambiguity exists, so I would recommend changing just those
places.

--

-- 
Bill McQuillan <McQuilWP <at> pobox.com>

Tony Hansen | 26 Sep 2008 16:04

Picon

Re: On ABNF in 5322: Advice needed


I'm with Bill here: leave the document alone except where ambiguities
have been pointed out by the RFC editor. There aren't that many uses of
the word "phrase" in the document.

	Tony

Bill McQuillan wrote:
> 
> On Thu, 2008-09-25, Pete Resnick wrote:
> 
>> 5322 (nee 2822upd) is in AUTH48. During this process, the editor 
>> noted some confusion about some of the text: Sometimes when I say 
>> (e.g.) "phrase", it's not clear whether I'm referring to the ABNF 
>> terminal "phrase" or a description of it. It's somewhat easy to 
>> distinguish, for instance, "a quoted string" from "quoted-string", 
>> but other places it's trickier.
> 
> Quoting from RFC 5234, "Augmented BNF for Syntax Specifications", 2.1:
> 
>    Unlike original BNF, angle brackets ("<", ">") are not required.
>    However, angle brackets may be used around a rule name whenever their
>    presence facilitates in discerning the use of a rule name.  This is
>    typically restricted to rule name references in free-form prose[...]
> 
> This seems to me to allow angle brackets to be used in only those places
> where it is needed to remove ambiguity.
> 
> If I understand you correctly, you have already found a small number of
> places where ambiguity exists, so I would recommend changing just those

(Continue reading)

Arnt Gulbrandsen | 26 Sep 2008 10:09

Re: On ABNF in 5322: Advice needed

Pete Resnick writes:
> 5322 (nee 2822upd) is in AUTH48. During this process, the editor noted 
> some confusion about some of the text: Sometimes when I say (e.g.) 
> "phrase", it's not clear whether I'm referring to the ABNF terminal 
> "phrase" or a description of it.
> ...
> Choice (A): "Don't touch a damn thing, Pete! It's too much of a risk 
> to screw things up, and we're moving to Draft Standard. And nothing 
> is so ambiguous as to make any difference."

My preference would be a variation of (a). Don't touch anything except 
where the editor noted. Consider rephrasing those specific sentences, 
and glance fleetingly to see if the same sentences appear elsewhere in 
the document.

If this were a substantial problem in the audience's eyes, you'd have 
known about it long ago.

Arnt

ned+ietf-822 | 26 Sep 2008 16:08

Re: On ABNF in 5322: Advice needed


> Pete Resnick writes:
> > 5322 (nee 2822upd) is in AUTH48. During this process, the editor noted
> > some confusion about some of the text: Sometimes when I say (e.g.)
> > "phrase", it's not clear whether I'm referring to the ABNF terminal
> > "phrase" or a description of it.
> > ...
> > Choice (A): "Don't touch a damn thing, Pete! It's too much of a risk
> > to screw things up, and we're moving to Draft Standard. And nothing
> > is so ambiguous as to make any difference."

> My preference would be a variation of (a). Don't touch anything except
> where the editor noted. Consider rephrasing those specific sentences,
> and glance fleetingly to see if the same sentences appear elsewhere in
> the document.

+1

				Ned

Dave CROCKER | 28 Sep 2008 00:54

Re: On ABNF in 5322: Advice needed


ned+ietf-822 <at> mrochek.com wrote:
>> My preference would be a variation of (a). Don't touch anything except
>> where the editor noted. Consider rephrasing those specific sentences,
>> and glance fleetingly to see if the same sentences appear elsewhere in
>> the document.
> 
> +1

Folks,

This means that the presence of angle-brackets will be inconsistent.  Is that 
really ok?

d/

--

-- 

   Dave Crocker
   Brandenburg InternetWorking
   bbiw.net

Tony Finch | 29 Sep 2008 20:07

Picon

Re: On ABNF in 5322: Advice needed


On Sat, 27 Sep 2008, Dave CROCKER wrote:
> ned+ietf-822 <at> mrochek.com wrote:
> >
> > > My preference would be a variation of (a). Don't touch anything except
> > > where the editor noted. Consider rephrasing those specific sentences,
> > > and glance fleetingly to see if the same sentences appear elsewhere in
> > > the document.
> >
> > +1

+1

> This means that the presence of angle-brackets will be inconsistent.
> Is that really ok?

I understand "rephrase" to mean "adjusting the wording to remove the
ambiguity" which does not imply simply adding angle brackets.

Tony.
--

-- 
f.anthony.n.finch  <dot <at> dotat.at>  http://dotat.at/
DOVER WIGHT PORTLAND PLYMOUTH: WEST 4 OR 5, INCREASING 5 TO 7, PERHAPS GALE 8
LATER. MODERATE, INCREASING ROUGH OR VERY ROUGH. SHOWERS THEN RAIN. MODERATE
OR GOOD, OCCASIONALLY POOR.

Dave CROCKER | 29 Sep 2008 20:29

Re: On ABNF in 5322: Advice needed

Tony Finch wrote:
> On Sat, 27 Sep 2008, Dave CROCKER wrote:
>> This means that the presence of angle-brackets will be inconsistent.
>> Is that really ok?
> 
> I understand "rephrase" to mean "adjusting the wording to remove the
> ambiguity" which does not imply simply adding angle brackets.

I, too, understood that the result would eliminate contextual ambiguity.

However my point is that the result means "sometimes use angles and sometimes 
don't, depending upon whether (the author thinks that) ambiguity is a concern.

That type of rule is fine for general prose, but maybe not so good for formal 
specifications?

d/

--

-- 

   Dave Crocker
   Brandenburg InternetWorking
   bbiw.net

Tony Finch | 29 Sep 2008 20:36

Picon

Re: On ABNF in 5322: Advice needed


On Mon, 29 Sep 2008, Dave CROCKER wrote:
>
> However my point is that the result means "sometimes use angles and sometimes
> don't, depending upon whether (the author thinks that) ambiguity is a concern.

What I was trying to say is that if the rephrasing is done right there is
no need for angle brackets.

Tony.
--

-- 
f.anthony.n.finch  <dot <at> dotat.at>  http://dotat.at/
SOUTHEAST ICELAND: NORTHERLY 5 TO 7. ROUGH, OCCASIONALLY VERY ROUGH. RAIN OR
SQUALLY SHOWERS. MODERATE OR GOOD.

Pete Resnick | 29 Sep 2008 21:32

Re: On ABNF in 5322: Advice needed

On 9/29/08 at 7:36 PM +0100, Tony Finch wrote:

>What I was trying to say is that if the rephrasing is done right 
>there is no need for angle brackets.

And it did indeed turn out that no angle brackets were necessary, and 
very little rewording was needed.

Everything is done except for the msg-id-whatever issue. More about 
that in a different message.

pr
--

-- 
Pete Resnick <http://www.qualcomm.com/~presnick/>
Qualcomm Incorporated - Direct phone: (858)651-4478, Fax: (858)651-1102

Charles Lindsey | 26 Sep 2008 16:14

Picon

Picon

Re: On ABNF in 5322: Advice needed


Pete Resnick <presnick <at> qualcomm.com> wrote

>The suggestion was made (and it's been made by others) that all ABNF 
>terminals should be surrounded with angle brackets, so it's easy to 
>distinguish "a phrase" from "<phrase>" and "a quoted string" from 
>"<quoted-string>". My argument against doing this has always been (1) 
>there are very few places in the text where it is ambiguous, and none 
>(AFAIK) where it makes a difference; and (2) going through and making 
>the change could result in errors. (It's a lot of occurrences to go 
>through.) Advice sought:

+1

It makes reading such documents much easier. We went through all the
USEFOR drafts a long time ago to put them into that format. It takes some
effort, but IMHO it is worth it.

>If you choose (B), I'd also like to hear your answer to the question: 
>Does this require cycling back through the IESG or (*gasp*) another 
>IETF last call?

I would hope not. But it sounds as if the RFC Editor is willing to play
without that.

--

-- 
Charles H. Lindsey ---------At Home, doing my own thing------------------------
Tel: +44 161 436 6131 Fax: +44 161 436 6133   Web: http://www.cs.man.ac.uk/~chl
Email: chl <at> clerew.man.ac.uk      Snail: 5 Clerewood Ave, CHEADLE, SK8 3JU, U.K.
PGP: 2C15F1A9      Fingerprint: 73 6D C2 51 93 A0 01 E7 65 E8 64 7E 14 A4 AB A5

(Continue reading)

Arnt Gulbrandsen | 27 Sep 2008 08:37

Re: On ABNF in 5322: Advice needed


Charles Lindsey writes:
> It takes some effort, but IMHO it is worth it.

IMNSHO, "takes some effort" is another way of saying "#$!$# <at> $ <at>  not in AUTH48".

Arnt

Abhijit Menon-Sen | 26 Sep 2008 10:06

Re: On ABNF in 5322: Advice needed


At 2008-09-25 17:26:26 -0500, presnick <at> qualcomm.com wrote:
>
> Choice (A): "Don't touch a damn thing, Pete! It's too much of a risk to 
> screw things up, and we're moving to Draft Standard. And nothing is so 
> ambiguous as to make any difference."

That describes my feelings exactly.

-- ams

Alessandro Vesely | 26 Sep 2008 08:47

Picon

Re: On ABNF in 5322: Advice needed

Pete Resnick wrote:
> The suggestion was made (and it's been made by others) that all ABNF 
> terminals should be surrounded with angle brackets, so it's easy to 
> distinguish "a phrase" from "<phrase>" and "a quoted string" from 
> "<quoted-string>".

As an occasional RFC reader, I haven't been able to spot the point 
where the angle brackets would be useful. If a passage requires that 
distinction, I would consider more instructive to explicitly explain 
it at the point where it is needed, rather than cluttering all the 
text with annoying angle brackets.

If all ABNF terminals were to be surrounded by angle brackets, then 
that would obviously be required within rules as well, thereby 
defining ABABFN (Angle Bracket Augmented Backus–Naur Form). RFC 5234 
says angle brackets are not required.

> Choice (A): "Don't touch a damn thing, Pete! It's too much of a risk to 
> screw things up, and we're moving to Draft Standard. And nothing is so 
> ambiguous as to make any difference."

+1

Return

Return to gmane.ietf.rfc822.

Project Web Page

Discussion of RFC 2822 ()

Search Archive

Language

Change language

Options

Current view: Threads only / Showing only 20 lines / Not hiding cited text.
Change to All messages, whole messages, or hide cited text.

Post a message
NNTP Newsgroup
Classic Gmane web interface
XML

XML

RSS Feed
List Information

About Gmane