[Top] [All Lists]

Re: [ietf-822] "--" first in lines to indicate start of signature

2016-01-04 13:23:41
Hi Russ,

On 1/2/2016 6:38 PM, Russ Allbery wrote:> "Murray S. Kucherawy" <superuser(_at_)gmail(_dot_)com> writes:

What other uses of "--" on a line by itself are there?  The only thing I
can think of is ASCII art.

I've written ASCII/ANSI/VT100 drivers and I don't recall a relationship other than some early body/text delimiter (ideal when you parse/read from the bottom, i.e. first delimiter encounter when reading up).

Some people use it as a break between several thoughts.  This has probably
grown since Markdown started using three hyphens to indicate a horizontal
rule.  The distinction between two hyphens and three hyphens is the sort
of thing that people don't always get right.

As I recall it, there was a "stripping war" and people tried to hide their tags using different identifiers. QWK was a popular mail reader and it used something like, if I recall the format:

  X  QWK brag line where X was an extended "block-looking" character.

and Sparky did it so that it didn't get recognized and stripped. QWK format didn't want to get involved any Fidonet "regulation." Overall, mailers can do what they want. A reader or mailer was going to add its tag no matter what so it was not uncommon to see a bunch of "tear/tag" lines lumped together.

Much of the IETF stuff came from Fidonet where we didn't have a HEADER block. We had kludge lines (Control A lines) that can appear anywhere in the payload. That was one of the beauties of going to a more flexible, text-structured RFC822 format where increasing hardware speeds made it more feasible. The Tear line and well as the Origin line were important body text delimiters since kludge lines were normally in the top and bottom of message.

  ^aOther kludges

   body text

   --- tear line
    ^APath Line(s)
    ^aSeen-by Nodes line(s)

So mail processor developers can imagine what would be the logical varying coding techniques. You can start at the bottom, get your seen-by and path lines, if any, and get to that first "tear line" or start at the top and get to the last one, etc.

That being said, the more common problem is software recognizing -- at the
start of a line as being the start of a signature even if it's followed by
other text, which runs into issues with em-dashes as represented in ASCII.

I don't like the idea of trying to standardize a "-- " tag although this character sequence seems to be popular among "some" IETF users, but not outside. As you said, other formats such as "--- " also exist and you have other variations as well. There should be some guidelines I think for MUAs to consider or document, but not fix them to any set, if we can avoid it.

Why dashes? Why not underscores? or Equal signs or in a modern extended character set world, Line (xC4) characters?

___ HLS
=== HLS
─── HLS

They all look "cool" and maybe using an extended code MAY help delimit it better, coding wise.


ietf-822 mailing list