timechanging NcML bug

[tlogan2000]

as noted in initial comment : #1 (comment) using the timeunitsChange="true" flag results in an offset of +5 days

Have been in contact with UNIDATA support and this is indeed a bug on their end. I am supposed to notified when a fix is available.

[huard]

ETA ? Is there an open ticket about this ?

[tlogan2000]

Not sure but they seem pretty quick... Sent the original ticket only tuesday. I can ask if they have an idea for the ETA. I guess it depends how complicated a fix it is for them. I have a feeling they might be assuming a gregorian calendar and we end up with offset days over time (my data was 'noleap' only by chance)

[tlogan2000]

Ticketing is flagged with a number but only in my email. I don't seem to be able to find a 'forum' on their support site

Ticket Details

Ticket ID: XYW-188900
Department: Support THREDDS
Priority: Normal
Status: Open

[huard]

Good enough.

ESGF considers that time units that vary over a simulation should be fixed by the data provider.

[tlogan2000]

Yep. I honestly have never come across a dataset that has changing time units but I guess we still want to validate this functionality

[sbiner]

J'ai eu ce bug moi aussi hier justement.

…

On Thu, Jan 9, 2020 at 10:05 AM Travis Logan ***@***.***> wrote: Yep. I honestly have never come across a dataset that has changing time units but I guess we still want to validate this functionality — You are receiving this because you were assigned. Reply to this email directly, view it on GitHub <#4?email_source=notifications&email_token=AD7CIHGS4HJKDIXOUEV5MMTQ444KHA5CNFSM4KEZFVWKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEIQTKOI#issuecomment-572601657>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AD7CIHGCSSCMEFBIEATLYK3Q444KHANCNFSM4KEZFVWA> .

-- ---------------------- Sébastien Biner ----------------------