r12a: Extra agenda item: document I sent.
MichaelM: Unissued doc describes architecture. Has not enough i18n.
Kaz: We talked about that in prev meeting already.
<addison> addison: notes that we didn't review the architecture document
McCool__: data format with readable string, has language, but not direction.
… We derive the direction from the language.
… json-ld should solve this, we think.
… default language sets direction.
… "title" and "description" appear in multiple places.
… They can have multiple strings each with a lang tag.
… When retrieving a thing description, might do HTTP negotiation.
… Allows for richer environment. Also need to markup default lang.
… Issue with lang nego is that you may end up with diff. lang than what support tag provides.
… So we decided that the support tag should always be a URI.
… We decided to use "title" instead of "name" and thus use the title mechanism.
addison: We read the spec and saw your responses to our comments.
… Maybe some nuances and some comments.
… I just submitted a list of comments on date/time.
<Zakim> r12a, you wanted to discuss 3 comments related to bidi
addison: Today lets discuss open issues.
r12a: First of all, thanks for the excellent self-review!
… The way you describe how to get to bidi info does not include first-strong heuristics.
<addison> https://github.com/w3c/wot-thing-description/issues/643
r12a: I think that should be the first heuristics.
McCool__: The strong-first is reasonable when there is a human to fix it if the heuristic fails.
… Not so much between machines.
addison: You send the string to something with the capacity to display it.
McCool__: It may be just be a thermostat that displays a string.
addison: If it has the capacity to display bidi, it will be able to do the heuristic.
<addison> https://w3c.github.io/wot-thing-description/#thing
r12a: You say there is a place to specify the default language. But you don't say there *must* be a default.
<taki> https://w3c.github.io/wot-thing-description/#thing
McCool__: We can add first-strong as an allowed first heuristic.
… We can say it can be applied in absence of lang info.
<taki> https://w3c.github.io/wot-thing-description/#thing
Taki: first-strong is already mentioned in the spec.
<McCool__> https://w3c.github.io/wot-thing-description/
taki: The big table.
addison: the table and the green box.
McCool__: Should we say first-strong should be used?
addison: yes.
McCool__: Is there an issue open for that already?
addison: your issue 643
<addison> https://github.com/w3c/wot-thing-description/issues/635
Looking at the issue on McCool__'s screen.
Text in bullet list needs updating.
r12a: second issue is when you have, say an arabic string that starts with html code. You'd have to label it as english to get the code to display, but actually it is arabic...
addison: why as english? you want rtl, don't you?
r12a: you might be right...
… In such a situation, the lang might be there. But if the text comes form a user filling a form, it may not.
… If the string can be edited, e.g.
addison: The cldr sub tags are meant to address it.
r12a: I think that is in the spec somewhere.
<taki> https://w3c.github.io/wot-thing-description/#titles-descriptions-serialization-json
McCool__: [looking through spec]
<addison> https://w3c.github.io/wot-thing-description/#thing
addison: section 6.3.2
McCool__: "default language... must include sub-tag..."
addison: People read these specs naively. Have no context. Better to point and not overspecify.
… Don't want people to write en-lat when there is really no need for subtags.
<addison> https://github.com/w3c/wot-thing-description/issues/635
r12a: It already says: "... *if* a lang can be written in more than one script".
addison: see my suggested text in ^^
<addison> TD Processors should be aware of certain special cases when processing bidirectional text. They should take care to use bidi isolation when presenting strings to users, particularly when embedding in surrounding text. Mixed direction text can occur in any language, even when the language is properly identified. TD producers should attempt to provide mixed direction strings in a way that can be displayed successfully by a naive user agent. For exa[CUT]
r12a: Other possible way is to just say json-ld hasn't solved it, we can't solve it at the moment.
McCool__: We may run into issues. Spec usage is one or two years out.
r12a: If you worry about first-strong algo in small devices, there is quite a lot more processing involved in detecting base direction from language tags.
r12a: Can mark it "at risk" as well, then you can pull it later.
addison: It is a recommendation for consumers: "you should...". That's good enough.
McCool__: We'll try to word it with two shoulds and try to avoid conflicts.
addison: Use lang info first, use first-strong in its absence.
r12a: In other words, you do first-strong unless you have more info, such as language info.
McCool__: We'll need to work on the "must" and "should".
addison: You can even use "may". You "should" use strong-first in the absence of lang info. You "may" use... etc.
<addison> https://github.com/w3c/wot-thing-description/issues/635#issuecomment-490545550
McCool__: We should still mention the cldr sub tag, but soften the language to "may".
<r12a> “In particular, producers of TDs should avoid numbers with embedded spaces in bidirectional text. Strings starting with embedded text using a script with a writing direction opposite to that of the base direction (for example, English words embedded in Arabic text) or with multidigit numbers should be avoided if possible.”
r12a: third isssue:
… [looking for link]
addison: I proposed text should be removed altogether.
r12a: The data needs to be what the data is, that’s why you have the other mechanisms described above. Rather than ‘avoid if possible’ say ‘use the directional override mechanism described above”.
addison: Don't want to tell people not to use bidi.
… We want people to exchange their actual text, not avoid it.
r12a: Don't say constrain your data, but do say watch out and use appropriate technology.
<JcK> For whatever it is worth, I've got the page display at full screen, at 150% magnification, on a 26 inch monitor. It may be merely a sign of my advanced age (in which case this is an accessibility complaint) but I still cannot read the text well enough to be able to understand what is going on as the cursor and text are moved around the screen.
McCool__: We warn people, because things are fragile and may break.
addison: They may display incorrectly, I wouldn't call that "break".
McCool__: We already have a green box about locale, with warnings.
addison: works for me.
McCool__: About date/time: may be just wording issue.
addison: People sometimes confused about offset vs time zone. They're not the same.
… Recommend to use UTC where you can.
McCool__: The device displaying it may have to localize it.
addison: That's a different thing.
… Carrying offsets just means adding/subtracting from the UTC.
JcK: Calendars are long-standing problem.
<JcK> Really, it is "user perception of specifying times and dates in the future" are a long-standing problem. No solutions or changes suggested, but be aware as you are writing the spec that the problem is out there and unsolved.
addison: 8601 didn't try to solve it.
… just use UTC for time stamps.
McCool__: I'll have to look at the text if it is clear about that.
taki: I mailed just before the meeting.
… Some things in the self-check list were not very clear.
<taki> The followings are items that we need advice from i18n group. https://github.com/w3c/wot-thing-description/issues/595 1. Reference BCP47 for language tag matching. need to add a description on language tag matching? https://github.com/w3c/wot-thing-description/issues/578 2. Allow for leap seconds in date and time data types. [Comment] Thing Description referes to dateTime in XSD Version 1.1 and it does not allow leap-se
<taki> conds. https://github.com/w3c/wot-thing-description/issues/585 3. Provide metadata constructs that can be used to indicate the base direction of any natural language string. 4. Specify that consumers of strings should use heuristics, preferably based on the Unicode Standard first-strong algorithm, to detect the base direction of a string except where metadata is provided.
<taki> https://github.com/w3c/wot-thing-description/issues/595 1. Reference BCP47 for language tag matching. need to add a description on language tag matching?
taki: See ^^
addison: Your spec successfully avoids specifying language matching. It mentions lang negotiaton but doesn't say how you do it.
… There are some possibilities.
McCool__: We mention HTTP negotiation, but don't give rules for it.
… We can probably leave that to the implementations. Won't break anything.
… I'll make a note about it.
<taki> https://github.com/w3c/wot-thing-description/issues/578
<taki> 2. Allow for leap seconds in date and time data types.
<taki> [Comment] Thing Description referes to dateTime in XSD Version 1.1 and it does not allow leap-seconds.
taki: Also question about leap seconds: there is no support for leap seconds?
addison: All that means is that seconds may go up to 60.
McCool__: We can easily support them.
… Can be used in certain cases only.
McCool__: All time stamps in the spec don't need nanosecond accuracy. But some payload may need accuracy.
kaz: One quick question: From previous call, I understand there are i18n issues with TD, but the architecture itself is quite abstract and maybe wouldn't require i18n review.
addison: Didn't look closely, but you'll never know...
McCool__ and Kaz discuss next steps and time line.
Adjourned.