#24 Bug(s) in transliteration of vocabularies
I am using the JSR-188 RI to parse profiles for mobile
devices.
There seems to be a bug in at least one of the
vocabularies.
Take the example of Nokia 6630 profile
(http://nds1.nds.nokia.com/uaprof/N6630r100.xml):
Setting up the framework in an obvious way we can
print out the values of the various attributes.
For the URL above, the value returned for CcppAccept
is "application/java" - a string.
For MmsCcppAccept, however a HashSet is returned
with a collection of MIME types.
Visual inspection of the profiles reveals:
(a) The values for *both* attributes is a Bag of MIME
types.
Yet one bag returns a String, the other, a HashSet.
(b) Comparison with the profiles for the respective
attributes
(http://www.wapforum.org/profiles/MMS/ccppschema-
20010111#MmsCharacteristics),
(http://www.openmobilealliance.org/tech/profiles/UAPROF
/ccppschema-20021212#SoftwarePlatform)
CcppAccept is defined thus:
<rdf:Description rdf:ID="CcppAccept">
<rdf:type
rdf:resource="http://www.w3.org/1999/02/22-rdf-
syntax-ns#Property" />
<rdf:type
rdf:resource="http://www.w3.org/1999/02/22-rdf-
syntax-ns#Bag" />
<rdfs:domain rdf:resource="#SoftwarePlatform" />
<rdfs:comment>Description: List of content types the
device supports. Property value is a list of MIME types,
where each item in the list is a content type
descriptor as specified by RFC 2045. Type: Literal (bag)
Resolution: Append
Examples: "text/html", "text/plain", "text/html", "image/gif
"</rdfs:comment>
</rdf:Description>
MmsCcppAccept is defined thus:
<rdf:Description ID="MmsCcppAccept">
<rdf:type
rdf:resource="http://www.w3.org/2000/01/rdf-
schema#Property" />
<rdf:type
rdf:resource="http://www.w3.org/2000/01/rdf-
schema#Bag" />
<rdfs:domain rdf:resource="#MmsCharacteristics" />
<rdfs:comment>Description: Property value is a list of
supported content types where each item in the list is a
content type name registered as a MIME type Type:
Literal bag Resolution: Locked
Examples: "image/jpeg", "audio/wav", "video/mpeg-
4"</rdfs:comment>
</rdf:Description>
Both are Bags.
Yet, in the *.xml files, we have:
<attribute-desc name="CcppAccept" base-
type="literal" composition="simple" resolution-
policy="append"/>
and
<attribute-desc name="MmsCcppAccept" base-
type="literal" composition="set" resolution-
policy="locked" />
This looks like a mistranslation for CcppAccept. When I
edit the xml, replacing "simple" with "set",
I get a HashSet of MIME type strings returned, which is
what I would expect.
This mistranslation is true for numerous other entries in
that xml file (only one I have inspected so far). All were
sets and sequences.
The potential impact of this kind of bug is potentially
extensive: there are 22 MIME types in this
devices 'accept' header, 77 in the CC/PP profile, 14 of
which are shared.
I'll be examining the other vocabulary transliterations,
will report results and submit patched versions of the xml
for review - this was just a bugfix, so timescale
uncertain because of other priorities.
I'm appending my patched version of the XML - usual
disclaimers as to accuracy etc etc.
Patched version of ccppschema-20021212.xml
Logged In: YES
user_id=1206260
I note that this has been raised before as bug 859161 which
has a closed status.