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

[Robert Wilton <[e-mail]>]

On 04/03/2019 11:49, Martin Bjorklund wrote:
> "Rob Wilton (rwilton)" <[e-mail]> wrote:
>> But this behaviour is still different from the frequently used meaning
>> of ‘\’ today in programming languages, which as I know it just splits
>> lines and preserves whitespace.
> Right, but we're not doing a programming language, we try to fit long
> lines in examples into the format required by RFCs, including
> indentation.

But this does give a different meaning to a well used notation, that may 
well cause surprise and confusion.


>    For example, suppose you have:
>
>    Here's an example:
>
>    foo bar baz \
>    buzz
>
> Unfolded, do you think this is:
>
>    foo bar baz buzz
>
> or
>
>    foo bar baz    buzz

This one.  The counter example:

   foo bar baz\
   buzz

Do you think that this is:

   foo bar bazbuzz

Or

   foo bar baz  buzz


>
>
>> YANG requires a separator character between a keyword and its
>> argument.  What happens if the tool happens to split the line between
>> the two of them.
>>
>>
>>
>> E.g
>>
>>                         container\
>>
>>    test
>>
>>
>>
>> After the artwork folding, this would become
>>
>>                         containertest
>>
>>
>>
>> This could be mitigated by a smarter folding tool (e.g. split before
>> the ‘r’ or after the first space).
>
> 1.  Don't use a tool to add line breaks - remember the goal was to
>      have a readable example.

No line break is added.  It was performing a line split on "container test".


>
> 2.  Don't use this alg. for YANG modules, since YANG has builtin /
>      native support for splitting long lines.
>
>
>> Or, what if the tool was being used to fold a table that was using
>> whitespace to align the columns.  This could easily break if
>> whitespace is stripped.
> Don't use this alg for tables (or ascii art in general) -- it won't
> help readers.

The choice of solution seems to make it quite restricted where it can be 
validly used.  I.e. it is only safe to use to fold documents where 
whitespace has no significance, and can be arbitrarily removed from the 
folded object without changing the document's meaning.

I think that this would mean that the folding mechanism isn't safe on 
either XML or JSON documents (since whitespace is significant, e.g. 
within a string).

There seems to be a ever growing list of things that we can't safely use 
this form of artwork folding on.  Are there are any examples of stuff 
that we can safely use it on? ;-)

Thanks,
Rob


>
>
> /martin
>
>
>
>
>
>>
>>
>> Thanks,
>> Rob
>>
>>
>>
>>
>>
>> -----Original Message-----
>> From: Martin Bjorklund <[e-mail]>
>> Sent: 04 March 2019 08:40
>> To: Rob Wilton (rwilton) <[e-mail]>
>> Cc: [e-mail]; [e-mail]; [e-mail]
>> Subject: Re: [netmod] artwork folding: dual support modes?
>>
>>
>>
>> "Rob Wilton (rwilton)" <[e-mail]<mailto:[e-mail]>>
>> wrote:
>>
>>> Hi Adrian,
>>> I mostly agree with your last sentence.
>>> I think that if you always preserve whitespace then a single slash is
>>> fine.  I.e. the single slash just breaks the line, and I think that
>>> this matches how editors, programming languages, etc normally behave.
>>> What I’m not keen on is using a single slash, and then automatically
>>> stripping leading whitespace on the line following a slash.
>>
>>
>> And this is the solution that I prefer.  The reason for this is that I
>> view examples as something that is there to illustrate some point to
>> the reader, and I think that a prettier formatted example with less
>> noise helps the reader.  I also think that is most cases, this works
>> well; i.e., most cases are _not_ on the form:
>>
>>
>>
>>     12345      78990
>>
>>            ^-------------- I need a line break here
>>
>>
>>
>>
>>
>> For this problem, I prefer a solution that is intuitive and has less
>> noise and works for 90% of the cases, than a less intuitive solution
>> with more noise that works for 100% of the cases.
>>
>>
>>
>>
>>
>> /martin
>>
>>
>>
>>
>>
>>
>>
>>> If we want to have control of layout and be able to strip extra
>>> whitespace then my argument is that it is better to be explicit, and
>>> using two slashes is one way of achieving this.
>>> Thanks,
>>> Rob
>>> From: netmod <[e-mail]<mailto:[e-mail]>>
>>> On Behalf Of Adrian Farrel
>>> Sent: 27 February 2019 09:41
>>> To: 'Joel Jaeggli' <[e-mail]<mailto:[e-mail]>>
>>> Cc: [e-mail]<mailto:[e-mail]>
>>> Subject: Re: [netmod] artwork folding: dual support modes?
>>> Complete agreement, Joel.
>>> What follows may look better in proportional fonts.
>>> With a single slash we can wrap as follows
>>> 1234567        9012345
>>> Goes to…
>>> 1234567    \
>>>      9012345
>>> …and unwrapping is easy.
>>> However, if I want to manually wrap the line with indentation
>>> The quick brown fox jumps over the lazy dog
>>> ..going to…
>>> The quick brown fox\
>>>        jumps over the lazy dog
>>> …I am going to unfold as…
>>> The quick brown fox      jumps over the lazy dog
>>> Conversely, if I resolve this second case by stripping leading spaces
>>> I get…
>>> The quick brown foxjumps over the lazy dog
>>> So I have to fold as…
>>> The quick brown fox \
>>>        jumps over the lazy dog
>>> But this causes the first case to unfold as
>>> 1234567    9012345
>>> …i.e., with missing spaces.
>>> This is what caused the use of the second slash so…
>>> 1234567    \
>>> \    9012345
>>> …and…
>>> The quick brown fox\
>>>       \ jumps over the lazy dog
>>> So, my point is, if and only if we do not care about these “spaces on
>>> the fold” cases, we can operate with a single slash.
>>> Cheers,
>>> Adrian
>>> From: Joel Jaeggli
>>> <[e-mail]<mailto:[e-mail]<mailto:[e-mail]%3cmailto:[e-mail]>>>
>>> Sent: 27 February 2019 06:31
>>> To: Adrian Farrel
>>> <[e-mail]<mailto:[e-mail]<mailto:[e-mail]%3cmailto:[e-mail]>>>
>>> Cc: Kent Watsen
>>> <[e-mail]<mailto:[e-mail]<mailto:[e-mail]%3cmailto:[e-mail]>>>;
>>> [e-mail]<mailto:[e-mail]<mailto:[e-mail]%3cmailto:[e-mail]>>
>>> Subject: Re: [netmod] artwork folding: dual support modes?
>>> On Feb 26, 2019, at 14:26, Adrian Farrel
>>> <[e-mail]<mailto:[e-mail]<mailto:[e-mail]%3cmailto:[e-mail]>>>
>>> wrote:
>>> Hey.
>>> I’ve been having this discussion with Kent off-line, but thought it
>>> should come to the list.
>>> I don’t think it is a good idea to have two approaches. While it would
>>> be relatively easy to code for both approaches, it seems to add a
>>> degree of confusion if both have to be handled by the same code
>>> (consider deciding whether leading space characters are to be retained
>>> or not, something that can only be decided when the first non-space
>>> character is found), or by having different code for the two different
>>> cases.
>>> It doesn’t seem to me that both cases are needed. We can pick one or
>>> the other.
>>> A single slash has been used to wrap long lines in editors and shells
>>> for decades at this point.
>>> and yeah whatever it is one method seems better than two.
>>> And *if* we want to allow manual folding so that indents can be made
>>> to make the document more human-readable then we have to use a leading
>>> ‘\’ on continuation lines to show which spaces should be stripped and
>>> which retained.
>>> Cheers,
>>> Adrian
>>> From: netmod
>>> <[e-mail]<mailto:[e-mail]<mailto:[e-mail]%3cmailto:[e-mail]>>>
>>> On Behalf Of Kent Watsen
>>> Sent: 25 February 2019 22:22
>>> To:
>>> [e-mail]<mailto:[e-mail]<mailto:[e-mail]%3cmailto:[e-mail]>>
>>> Subject: [netmod] artwork folding: dual support modes?
>>> I had a chat with the tools team recently and, in the course of
>>> things, it was implied that the double backslash approach we have now
>>> was both surprising and non-intuitive.
>>> This got me thinking that we may have thrown the proverbial baby out
>>> with the bathwater.
>>> That is, currently we have a header that reads:
>>>    NOTE: '\\' line wrapping per BCP XX (RFC XXXX)
>>> So why not *also* support a header that reads (note the singe slash):
>>>    NOTE: '\' line wrapping per BCP XX (RFC XXXX)
>>> Whereby this second form only supports the folded line continuing on
>>> column 1 (no indents).
>>> Thoughts?
>>> Kent // contributor
>>> _______________________________________________
>>> netmod mailing list
>>> [e-mail]<mailto:[e-mail]<mailto:[e-mail]%3cmailto:[e-mail]>>
>>> https://www.ietf.org/mailman/listinfo/netmod