And uses the new pagination scheme, instead of origin ids.
Why don't you add type here and not within the docstring?
I thought this was to be named count?
same remark than for in-memory implementation (type, name, typo).
Why do we force the string type here?
Why don't you change the db.origin_get_range implementation so that it retuns what you want instead?
Because I wrote that diff in mid-november and didn't touch it since; and we still used Python 3.5 at the time. I'll have to rebase it
Yeah I guess we're not consistent here. I have a small preference for limit, sorry for asking the change on the other diff, I didn't realize the inconsistency
Because it's an opaque token, so clients shouldn't know what it actually contains so we can change it at any time. (eg. for Cassandra, it will be a large byte array)
Because I need the origin id to compute the next_page_token.
Build has FAILED
Link to build: https://jenkins.softwareheritage.org/job/DSTO/job/tox/837/
See console output for more information: https://jenkins.softwareheritage.org/job/DSTO/job/tox/837/console