Re: [netmod] Last Call: <draft-ietf-netmod-artwork-folding-07.txt> (Handling Long Lines in Inclusions in Internet-Drafts and RFCs) to Best Current Practice

[Paul Kyzivat <[e-mail]>]

On 8/2/19 9:22 AM, Erik Auerswald wrote:
> Hi Paul,
> 
> I'll try to elaborate on my understanding of "forced folding" below:
> 
> On 02.08.19 02:08, Paul Kyzivat wrote:
>> On 7/31/19 3:41 PM, Kent Watsen wrote:
>>> [...]
>>>> 2) Regarding forced folding in section 8.2.1: step 1 (Determine 
>>>> where the fold will occur) could benefit from some elaboration 
>>>> regarding lines that require forced folding. In particular, when an 
>>>> input line flagged for forced folding ends in backslash, then it 
>>>> must indeed be folded before the last character, while in all other 
>>>> cases it can be folded at any position prior to the max line length. 
>>>> (While this is obvious if you think about it, some might miss this.)
> 
> If the original text data contains the folding sequence, one line of
> that text data ends in a backslash. If such a line is folded before
> the backslash, the folded data still contains the folding sequence that
> came from the original text data (in addition to the folding sequences
> inserted by folding).
> 
> As I see it, forced folding would insert a new folding sequence inside
> of the original folding sequence.
> 
> Example before folding:
> 
> foo\
>    \bar
> 
> Example after folding:
> 
> == NOTE: '\\' line wrapping per BCP XX (RFC XXXX) ==
> 
> foo\\
> \
>    \bar

OK. I was still thinking of it a bit wrong.



> A line that requires forced folding and that is already of maximum
> length would be lengthened by forced folding, thus it would need to
> be folded twice.
> 
> This, of course, shows that unfolding must not recursively try to
> unfold an already unfolded part of a line again.
> 
>> [...]
>> (Or do you want to allow the degenerate case where the line is folded 
>> at the end, so the continuation contains nothing?)
> 
> I'd say that this degenerate case is required for forced folding to
> work.

Yeah, it seems so. It would be silly to do in any other case, but I 
guess it is not forbidden.

But now that I am thinking about this, it occurs to me that maybe the 
two methods can be collapsed into one:

just make the sequence to be inserted for folding be (using ABNF):

"\" %x0C [%x0A] *" " ["\"]

The trailing backslash only needs to be inserted if the character 
following it is a space or backslash. (IOW it is *optional* if the net 
character is *not* space or backslash.)

So, when removing the folding, upon finding ("\" %x0C [%x0A] *" ") you 
remove that and if there is then a backslash remove that too.

	Thanks,
	Paul