Closed lars-hellstrom closed 7 years ago
the wording of the standard explicitly says href and says it takes a URI see the description of references and the example in figure 3.1, looks like the example in 3.1.3.1 got missed (this attribute oscillated between href and xref in earlier drafts as I recall)
Actually, the RNC in 3.1.1 and the "informal description" in 3.1.2 use href
consistently and a #
in the fragment identifier , and then 3.1.3 uses xref
and no #
consistently in all examples and the text.
What a mess. I think we should just stick with xref
and #
and clean up 3.1.3.
I will do that and make a pull request.
cc: @JamesHDavenport
@kohlhase no I think the intended scheme is to use href and a uri which is used everywhere except for 6 bad instances of "xref" in the spec.
see also all uses of OMR in the CDs which all consistently use href
$ grep -o '<OMR[^<>]*' */*/*ocd
cd/experimental/field3.ocd:<OMR href="#pr"/
cd/experimental/field3.ocd:<OMR href="#prn"/
cd/experimental/field4.ocd:<OMR href="#pr"/
cd/experimental/field4.ocd:<OMR href="#pr"/
cd/experimental/linalgeig1.ocd:<OMR href="#pr"/
cd/experimental/linalgeig1.ocd:<OMR href="qr"/
cd/experimental/linalgeig2.ocd:<OMR href="#ev"/
cd/experimental/polynomial3.ocd:<OMR href="#pr"/
cd/experimental/polynomial3.ocd:<OMR href="#r"/
cd/experimental/ring3.ocd:<OMR href="#domain"/
cd/experimental/ringname1.ocd:<OMR href="#fr"/
cd/Official/scscp1.ocd:<OMR href="scscp://somehost:26133/92cfceb39d57d914ed8b14d0e37643de0797ae56"/
cd/Official/scscp1.ocd:<OMR href="scscp://somehost.somedomain:26133/q9t4eX"/
cd/Official/scscp2.ocd:<OMR href="scscp://somewhere:26133/qjhtkoeurthoedu"/
cd/Official/scscp2.ocd:<OMR href="scscp://somewhere:26133/qjhtkoeurthoedu"/
cd/Official/scscp2.ocd:<OMR href="scscp://somewhere:26133/qjhtkoeurthoedu"/
What a mess. I think we should just stick with xref and # and clean up 3.1.3.
Oh, I am still contributing to the mess, I wanted to say "stick with href
and ...."
@kohlhase or did you mean "stick with href and #
?
Ah OK then yes agreed:-)
yes, see PR #12
closing here in favor of #12.
Section 3.1.3.1 of the standard uses
xref
(with no#
before the fragment identifier) as the attribute ofOMR
. The RNG schema says it should behref
, and that the value is an URI.This was previously reported in: OpenMath/OM3#104