javamonn / sunspot Goto Github PK

• default to including bundles (for backwards compat)
• maybe first option in "meta" section

• diff between created time and send complete time
• diff between create time and web push receipt in service worker
• count of alerts sent

searching by collection name returns lots of spam collections - there should be an easy way to sort on "activity" and filter out collections where theres an order of magnitude difference, otherwise sort on volume

• filter by name (configurable modifier, e.g. "includes" / "starts with" etc or support regex)?
• filter by token ids

now that the oauth integration is created on load, the wallet connect step of the oauth create flow can be deferred to the time of alert creation, similar to the primary alert create flow

use for "snoozing" delivery

silent in the api: https://developer.mozilla.org/en-US/docs/Web/API/ServiceWorkerRegistration/showNotification

it seems like metadata isnt always pulling correctly for this collection

image-notification-url redirects to the source url, but the source sets a content-disposition header to force download, which discord / slack fails to render - i think restoring functionality to download and return directly perhaps makes more sense

source image url: https://antonymnft.s3.us-west-1.amazonaws.com/4848.png

currently backend treats this as a destination missing access error - should set a more specific disabled reason, surface in client, and only disable the offending alert if multiple embeds are batched

• add pwa manifest + prompt install
• optimize mobile layout
• swap hand rolled wallet connect out for https://github.com/tmm/wagmi + ethers
• consider push notification experience (support multiple destinations per alert or update web push to support multiple endpoints)

when many alerts are created, it becomes difficult to manage.

potential solutions:

• add more methods of distinction on the top level page (i.e. include destination)
• add filtering on the different columns (filter by collection, filter by destination, etc)
• add something like "watched collections" (becomes even more relevant with collection wide alerts like price / volume notifications)

• link to collection with alert filter from list view
• add link to overflow menu in alert detail
• add links when filtering collections in new / update alert (to better filter out scam collections)

currently, numerical traits are filtered in the same way as string traits, where as a relative filter ("trait < / > / = value") would be more useful

probably check box in left col with an "action" dropdown in the top right

• bulk enable
• bulk delete

the web app will sometimes seem to hang after periods of inactivity, though im able to reproduce this consistently. i would imagine this has more to do with the background worker, but perhaps profiling will help surface the problem

the copy of notifications should be customizable and interpolate variables contextually.

requests in discord:

• https://discord.com/channels/894608871539228754/934103196928978977/935702111570784296

searching by contract address (https://etherscan.io/address/0xa78e4da68b26adafb179618573dd0a9fc79233c1) for karafuru (https://opensea.io/collection/karafuru) returns a stale slug (karafuru-v2-1)

twitter v2 api has route specific limits relative to the number of authorized accounts, and its possible for a single bot reporting at a high frequency (e.g. a sales pump) to "overflow" the limit derived just from its authorization, and start eating into limits that should be provisioned for other bots.

• apply for increased api tier
• book-keep account specific limit + exhaustion and stop creating tweets when exhausted
• consider throttling + batching strategies during periods of high activity

i wonder if the websocket is getting timed out - look into discord.js config, or automatically roll the pod on some schedule in k8s