On Fri, 24 Oct 1997, Glenn Adams wrote:
> I don't see any problem in ISO-2022-JP with conformance to ISO-2022;
> the only questionable use may be that it implies that the G0 designation
> is reset at MIME line boundary.
It definitely does not "imply" this, in the sense that you do not
have to do it, but it's done anyway; it requires that the ESC
sequence to do this is present at the end of the line.
> This is not really a problem w.r.t.
> MIME processing, as it is simply saying that conversion should occur
> on the granularity of individual lines. This consititutes an application
> layer protocol on top of ISO-2022. Of course this does pose some problems
> for treating ISO-2022-JP in non-MIME contexts or quasi-MIME contexts
> that don't adhere to MIME CRLF rules (like HTTP). I think actual practice
> ignores this whole issue anyway and simply does not reset the G0
> designation at all. I recently spoke with Mark Crispin about his
> implementation, and he does not do such resets.
Glenn - I am currently not sure what it is that does not conform.
RFC 1468 cleary requires that at the end of the line, there is a
switch back to ASCII if we are not already in ASCII. This is no
problem for ISO 2022, as ISO 2022 does not prohibit to switch
back and forth between various things without every using them.
I suspected that the problem is that iso-2022-jp uses designations
but not invocations, but 6.2.4 of ISO 2022 (in my 1986 version)
clearly says that:
When a graphic set is designated by an escape sequence, and if
that class of graphic set (i.e. G0, G1, G2 or G3) is currently in-
voked, then the new set shall also be invoked.
I will contact Prof. Shibano to investigate why JIS X 0208:1997 says
that iso-2022-jp does not conform to ISO 2022.
Regards, Martin.
This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:37 EDT