[rfc-i] diagram issues

Paul Kyzivat pkyzivat at alum.mit.edu
Wed Jun 25 09:21:51 PDT 2014

On 6/25/14 11:48 AM, Joe Touch wrote:
> On Jun 25, 2014, at 8:31 AM, Paul Kyzivat <pkyzivat at alum.mit.edu> wrote:
>> On 6/24/14 9:50 PM, John Levine wrote:
>>>>> I'm particularly concerned about the notion that figures that aren't
>>>>> available as ASCII-art will be resolved using URLs.
>>>> Do you have a better suggestion?
>>> I'd think that if someone deeply cared about this, it wouldn't be hard
>>> for him to write a postprocessor that that fetches the image URL,
>>> renders the SVG into a bitmap using one of the many available
>>> libraries, then passes that through something like ppmascii to turn
>>> the bitmap into ASCII art and stick that into the text file.
>> The fundamental problem is resolution. You can make a drawing with a lot of fine lines, differing line styles, labels with small fonts, and labels written vertically, that you cannot render in ascii art.
> It is difficult to replicate a Van Gogh in ASCII art, but IMO diagrams that are too complex for one page of ASCII art are also probably not useful as protocol diagrams either.

I disagree. Lots of things are too complex to render in an ASCII art 
diagram. The usual work-around is factoring - break a single diagram 
into several smaller diagrams. This is sometimes a good thing. But other 
times it simply makes the relationship between pieces harder to understand.

The idea that "a picture is worth a thousand words" depends upon the 
picture having enough detail to represent a thousand words. IMO an ASCII 
art picture is only worth a hundred words.

> I fear that SVG will be an excuse for illegible diagrams.

Look at the ascii art diagrams in existing RFCs. Plenty of them are 
pretty illegible, and would be more legible with SVG.

If the SVG diagrams are illegible then that can be solved with better 
reviews. When ASCII art diagrams are illegible because of limitations in 
what can be done with ASCII art then reviews are less likely to solve 
the problem.

IMO a more significant concern is that any sort of diagram (even ASCII 
art) that is *generated* by a tool from some other original 
representation is problematic if the original representation isn't 
archived along with the document. This isn't problematic for the meaning 
of the standard, but it is problematic for future revisions. It is less 
of a problem for RFCs, since bis's are infrequent, but it is a big 
problem for drafts.



More information about the rfc-interest mailing list