provide/adjust dandiset stats/description/url for a dandiset upon push

dandi / dandisets

738 Dandisets, 811.8 TB total. DataLad super-dataset of all Dandisets from https://github.com/dandisets

10 stars 0 forks source link

provide/adjust dandiset stats/description/url for a dandiset upon push #20

Closed yarikoptic closed 3 years ago

yarikoptic commented 3 years ago

ATM those on https://github.com/dandisets/ have no metadata on them. Hard to impossible to find anything unless know which one you are looking for. Ideally within https://github.com/dandi/dandisets/blob/master/tools/backups2datalad.py we should

upon initial publish for each repo provide URL which would point to https://identifiers.org/DANDI:<ID> e.g. https://identifiers.org/DANDI:000006 .
upon every push update description so it becomes <files unicode> #files <storage unicode> size, Contact, Description to mirror what we have on https://gui.dandiarchive.org/#/dandiset .
there should be a way to "force trigger" update of descriptions/url (e.g. to populate currently already pushed to github datasets or later if we change what we display there etc)

jwodder commented 3 years ago

@yarikoptic The "file" and "storage" symbols on gui.dandiarchive.org appear to be part of a custom font rather than actual Unicode characters. There are some Unicode characters similar to the "file" glyph, but they don't show up on my computer. I can't find a Unicode character resembling the "storage" glyph.

jwodder commented 3 years ago

@yarikoptic What exactly should the semantics of the "force trigger" option be? Should it cause the script to run as normal but also set the description & URL for every repository, regardless of whether they're being pushed to, or should it cause the script to only set the description & URL for repositories, not even bothering with syncing?

What if the script always set the description & URL for every repository operated on regardless of whether they're pushed to?

yarikoptic commented 3 years ago

storage: ⛃ (well -- for https://en.wikipedia.org/wiki/Draughts but looks close enough), file -- I failed to find one. Which one you found? Oh well -- could be X files, Y GBs without fancy unicode.

... set the description & URL for repositories, not even bothering with syncing?

I think it would be useful to support such a dedicated command

What if the script always set the description & URL for every repository operated on regardless of whether they're pushed to?

would be wasteful/take time as number of dandisets grows. Those could change only whenever a dandiset is updated (we would be pushing) or we decide to change something (formatting etc), which would be triggered manually via a dedicated command