ipfs-webui

ipfs-webui copied to clipboard

Right now we do pin.remote.add with background:true.

It's fine for smaller files which get pinned pretty fast, but not really honest when it comes to bigger files.

We should (rough draft)

• track pin status
  • this should not be blocking, keep background:false
  • store pins "in progress" in local storage as {service}:{cid}
    • this way we persist them across browsing sessions / laptop restarts etc
  • periodically check this storage and poll for status and update UI to reflect changes
    • avoid spamming remote service
      • interval should be based on time since pin event.
      • after each passing 30s we should increase interval and back-off to something more sensible (eg. 10s → 30s →1m → 5m → 15m)
• mark remote pin as "in progress" and indicate that in UI (could be "breathing" effect or something)
  • if "ongoing" pin fails we should
    • mark it as "failed" and indicate that in UI (icon tinted in red?)
    • clicking on failed cloud pin should
      • clear the 'failed' indicator
      • open pin modal with failed item unchecked
        • in the modal we could indicate which service failed with some icon, if possible, but after closing the modal that indicator goes away (one time thing)
• we should store "ongoing" pins in localStorage, so even when app is closes and opened later, we keep track of pins initiated from the webui
• when app starts it should restore any "ongoing" pins
  • read service:cid pairings from local storage
  • check states from each service via pin remote ls --service {service} --cid {cid1},{cid2} --status=queued,pinning,pinned,failed (where "cid1" and "cid2" are ongoing pins on "service")
  • update UI to reflect state
    • if pinned, remove from local storage
    • if queued/pinning, show the "ongoing" status and keep polling until its "pinned" or "failed"
    • if failed, mark it with the same 'failed' indicator

We should do the same tracking for local pins:

• node could be stopped before pinning finishes, so we need store them and resume any stale ones on startup
• removing "ongoing local pin" flag could be done at the end of local pin operation or via periodic check if pin exists
• there should be a broom (tm) that allows user to clean up entries for finished pin jobs