On 29.11.2019 23:18, John C Klensin wrote:
... I think others have commented on the assumption that pagination is good only if the document is headed for a line printer, but it may be worth pointing out that a recent errata submission may have identified a weakness in rfcdiff. An erratum submitted last April complained about an extended example in RFC 5321 being internally inconsistent about leading spaces that turn out to be important. (see https://www.rfc-editor.org/errata/eid5711 for details, but they aren't important). The problem is that the XML source appears to be correct -- leading spaces in both parts of the example -- but the XML output (under v2) isn't.
So what caused the problem in the XML output? (my copy of rfc5321.xml from AUTH48 seems to be truncated, so I can't tell).
This wasn't caught caught during AUTH48 or earlier, or apparently in the decade between 5321's publication and the erratum, at least in because the author was relying on rfcdiff and it is apparently insensitive to leading, as well as other types of spaces. I don't think that is a case for an emergency
That's a good reminder. I had a similar issue in the past, which I detected when doing an *actual* diff instead of rfcdiff. It's easy to forget though.
fix to rfcdiff, especially one that might have side-effects I haven't thought about, nor a fix to v2 of xml2rfc, but it should be another reminder than we two be very careful about what xml2rfc produces, especially in a world in which the XML is the canonical form but that there is no requirement that the IETF's favorite xml2rfc processor be used to process that source (and I don't know enough to know whether the IETF-approved Style Sheet covers this situation or not).
Well, we'd have to know about the actual cause. If it was an HTAB (as suggested by Carsten), then the answer is: - rfc2629.xslt added a warning back in 2014 - RFC 7749 warns about HTAB in <https://greenbytes.de/tech/webdav/rfc7749.html#element.artwork>: "Note that processors differ in the handling of horizontal TAB characters (some expand them, some treat them as single spaces), and thus these ought to be avoided." - RFC 7791 makes it an error: <https://greenbytes.de/tech/webdav/rfc7991.html#element.artwork>: "Tab characters (U+0009) inside of this element are prohibited." - I'm not sure what xml2rfc does in the current version, but I reported a bug back in April (<https://trac.tools.ietf.org/tools/xml2rfc/trac/ticket/403>) Finally, what do you mean by "IETF-approved Style Sheet"? Best regards, Julian
