Closed jmetz closed 4 months ago
yeah, I thought of a collection_staged.json
before.. (note the different name! 😕) but yeah... I can add that to backoffice 👍
Even better - collection_staged.json
probably makes more sense.
I think it should probably be updated on each stage
as well as publish
call, right?
Presumably more or less a copy/paste of the code you already wrote to create collection.json
, except that you'll pick out staged instead of published versions.
publish is not changing the staged versions (well, their status, but that is not included in the collection.json)
I added the dois to the collection.json files which are updated by backup
. so after backup
both collections are regenerated.
and yes after stage
staged collection and after publish the regular/published collection...
impl in 23ec0fb88b804718e42ebe10e91a4f19a5b25cd1
(sorry, developing this on main due to changes to gh wfs... ) 🤫
Part of https://github.com/bioimage-io/bioimageio-uploader-service/issues/9 (which in turn is part of meta-issue https://github.com/bioimage-io/bioimageio-uploader/issues/65 ) requires that we get all staged versions.
As this involves doing a list operation on S3 followed by collating all the individual
versions.json
files, and in principle this will happen many times between each new stage, I would recommend that we build this in backoffice whenstage
andpublish
are called, similar to the way thatcollection.json
is currently generated.What do you think @oeway + @FynnBe ?
The alternative, to have the
hypha
service dynamically generate the same JSON response every time the status page is loaded, seems very inefficient.