h-event time test should not normalise timezone -v2

[kylewm]

in microformats-v2/h-event/time.html

<span class="h-event">
    <span class="p-name">The 4th Microformat party</span> will be on 
    <ul>
        <li class="dt-start">
            <time class="value" datetime="2009-06-26">26 July</time>, from
            <time class="value">19:00:00-08:00</time> 
        </li>
        <li class="dt-start">
            <time class="value" datetime="2009-06-26">26 July</time>, from
            <time class="value">19:00:00-0800</time> 
        </li>
...

the output reads

"start": ["2009-06-26T19:00:00-0800", "2009-06-26T19:00:00-0800"...

with the colon removed from the first timezone. There is an issue discussing normalisation, but for now http://microformats.org/wiki/value-class-pattern#Date_and_time_parsing seems to indicate the colon should be preserved.

[glennjones]

Yes your right. My parser is been a bit too clever and normalising dates. I am using ISO 8601 date "profiles" which are like a set of sub rules that restrict the ISO 8601. The default profile drops colon from timezone.

The tests needs changing and I need to add a flag for "date normalisation" which is switch off by default. Unless of cause we add date normalisation to parsing rules : )

I will do this in next round of updates

[kylewm]

Normalizing is a proposal, I have it implemented and commented out in
mf2py. Happy to pursue that route instead if you prefer

[glennjones]

The tests now do NOT normalising dates as per current parsing rules. Leave this issue open while normalising dates is discussed.

[willnorris]

@tantek recently resolved http://microformats.org/wiki/microformats2-parsing-issues#parsing_a_dt-_property. It seemed to be mainly focused on normalizing dates with regards to space vs "T" as a separator and removal of implying 00 seconds. However, the vcp page is also now pretty clear that a resulting dt value should not have colons in the timezone.

[dissolve]

h-event/dates example is normalizing to add the colons still.

[gRegorLove]

Parsing for h-event/dates.html test doesn't use VCP, so it's expected to return the datetime attribute without normalization. There's a discussion about changing that (microformats/microformats2-parsing#12), but until that's resolved I think we can close this issue.

Edit: realized the OP is about a different test; my comment here was only about previous comment for h-event/dates

[jalcine]

Nudging this because this test is the only v2 test that's failing for the Rust library. Should we opt to keep (current behavior expected in the Rust library), preserve as authored, or drop the colon in offsets?

(Originally published at: https://jacky.wtf/2022/6/kduV)

[gRegorLove]

Specification updated (microformats/microformats2-parsing#56) to normalize by removing colon in timezone offset, so these tests can be updated accordingly. Note this only applies to VCP for now; see separate issue microformats/microformats2-parsing#12

[gRegorLove]

I reviewed the master branch and confirmed that 17d31db normalized the test outputs, so no additional test updates are needed to resolve this issue.