Re: [netmod] artwork folding: dual support modes?

[Kent Watsen <[e-mail]>]

> On Mar 4, 2019, at 11:04 AM, Martin Bjorklund <[e-mail]> wrote:
> 
> Kent Watsen <[e-mail]> wrote:
>> 
>> 
>>> But note that figures in RFCs are normally indented with 3 spaces
>>> (they _can_ be outdented, if the lines are long enough).
>> 
>> 
>> The days of scraping from plain-text RFCs are over [1].  Extracting,
>> if needed at all, should be from the XML, where there are no such
>> issues. Extracting from the plain-text output makes about as much
>> sense as extracting from the HTML or PDF outputs.
> 
> I am confused.  Are you saying that the unfolding algorithm only is
> supposed to work on data extracted from the XML version of the I-D or
> RFC?  If so, I think this needs to be clarified in the draft.

The unfolding algorithm works as long as the input == the output.  The 
problem is that plain-text RFCs introduce a lot of artifacts that makes 
lossless extraction difficult.  I don't believe we should try to design a 
solution for input != output.

Now that IETF has officially moved to XML as the sole format, there
is no longer a need to support extracting from plain-text.   In general, 
folks are advised to always extract from XML.   I support adding a 
statement to this affect.



>> Lossless extractions are critical for formal verifications (e.g.,
>> doctor reviews, shepherd reviews, AUTH48 reviews).  Both the
>> double-backslash approach we currently have, and the single-backslash
>> approach we had originally (where the continuation-line begins on
>> column 1, as it has been in programming languages for decades) provide
>> lossless extractions.
> 
> ... as does the single-backslash with leading space removal.

No, there are cases where this fails.  We went thru this before.  This is why
we adopted the double-backslash approach.


Kent // contributor  (also on my previous emails in this thread)