[rfc-i] sourcecode indentation

[tony at att.com (HANSEN, TONY L)]

On 2/13/16, 5:00 PM, "rfc-interest on behalf of Paul Kyzivat" <rfc-interest-bounces at rfc-editor.org on behalf of pkyzivat at alum.mit.edu> wrote:


>On 2/13/16 4:40 PM, Brian E Carpenter wrote:
>> On 14/02/2016 09:56, Paul Hoffman wrote:
>>> On 12 Feb 2016, at 15:52, HANSEN, TONY L wrote:
>>>
>>>> A better example of outdent would be
>>>>
>>>> <sourcecode indent="-4">
>>>>   foo
>>>>     bar
>>>>       baz
>>>> </sourcecode>
>>>>
>>>> which would generate:
>>>>
>>>> <pre class="sourcecode">
>>>> foo
>>>> bar
>>>>   baz
>>>> </pre>
>>>>
>>>> in the HTML format.
>>>>
>>>> A more common variation might be where the input is coming from elsewhere and you have no control over the indentation used
>>>> there:
>>>>
>>>> <sourcecode indent="-4" src="http://URL to the source code"/>
>>>>   or
>>>>
>>>> <sourcecode indent="16" src="http://URL to the source code"/>
>>>>
>>>> The use case for this is very specific, but probably not something that would get used often.
>>>>
>>>> If you think you would find it useful, please speak up.
>>>
>>> I will speak up against this idea. For sourcecode where indentation is important, seeing an example at the left margin might
>>> cause the reader to think that the code is in fact not indented. Your example above is a classic one: someone seeing that text
>>> would assume that "foo" and "bar" are at the same level when they are not. I would prefer not to have the RFC format have a
>>> display hinting mechanism that could be used to make examples harder to interpret.
>>
>> It's worse than that. In Python, it can invalidate the example completely.
>>
>> if whatever:
>>      this
>> that
>>
>> is very different from
>>
>> if whatever:
>>      this
>>      that
>>
>> but indent="-4" would reduce them both to
>>
>> if whatever:
>> this
>> that
>>
>> which is invalid syntax. It's a dangerous toy, in other words.
>
>Agreed that is dangerous. But negative indent could be OK if it just 
>generates an error if there isn't sufficient whitespace to do the 
>outdent completely.

I would actually prefer an error too if there is insufficient whitespace to do the outdent completely.

This,

  If whatever:
	This

Would generate an error for any indent < -2.


>But what does negative indent do if there are tabs at the beginning of
>the text?


This question is broader than indent and outdent, as we don't currently allow tabs anywhere in the XML source.

	Tony Hansen