1446
- Requires an accompanying commit in `odysee-frontend`.
- The change assumes that the `odysee-frontend` is the only project that uses these files directly, i.e. other clients will use the API instead.
- consolidate cases that need to auto set an active channel (like edit page) into channelSelector component
- also for consistency since some components would do it with button click and others on page mount
- prevent clear function on those pages (kind of a manual process to insert each page into the router condition)
From the previous commit, we are now showing toasts in the reverse order (latest to oldest).
Next, extend the "hide snack" timer to handle multiple snacks. It will dismiss them one by one, restarting itself until no more toasts.
Show a stacked GUI when there are multiple toasts.
Users can still manually dismiss the toasts.
Several issues with the clamping behavior:
- Problems trying to sync with the activeChannel setting.
- Corner-cases like unable to un-react because the comment and react channel was different; global mods not be able to change channels to do certain actions.
Just let the user know what are the channel(s) that they used to comment previously in the Toast.
If we can't confirm that user has not used another channel to comment, then don't allow commenting at all. Otherwise, there's no point doing the check.
## Why
Lost in a rebase. The GUI clamping was just meant to complement this (so that user knows what's going on), but ended up being the only thing that was pushed.
## Changes
- If the fetch failed or still in progress, we ask the user to wait or simply refresh the page.
- If already commented from another channel, bail and inform via Toast.
- this way, you can choose an active channel for something like commenting but when that component is unmounted, you always go back to the default channel (like leaving page)
- Moved from reach/ui to material/ui menu components, because reach ui wouldn't work with 2 menus
- This channel selector stores the default on settings
- setActiveChannelIfNotSet was deprecated, if the account has channels, it will always return a channel even if there is no active channel or stored channel
We were already limiting the call to once-per-5-minutes (for the succesful case), so just need to put failure cases into the mix.
Retry 3 times before stopping.
- Corrected flow definitions.
- Properly differentiate between "not yet fetched" and "no membership" as "undefined" and "<empty string>", respectively. There are GUI elements that need to know the unfetched case.
## Issue
1075, spam prevention.
## Approach
- When making a reaction, fetch reactions for all my channels for the particular comment id, and reject the reaction is any was found.
- Report the channel name in the toast so that user can at least know which channel to select in order to undo the reaction.
Completely remove any need to update things on our side when a Category is added or changed.
Will need to inform homepage owners to directly translate the 'label' field, so we don't need to use our translation system.
## Issue
??
## Behavioral Changes
- Use `claim_search` instead of `claim_list` to retrieve all all own claims with the same name (case-insensitive).
- Caveat: annonymous posts will be excluded.
- When a clash occurs, there is a possibility that we have multiple existing entries (e.g. "xxX", "xXx"). Since we don't know which one is best to fall back, I removed the "edit" button for this and replaced with a simpler text
## Code Note
- If not mistaken, the rest of the code still needs `selectMyClaimForUri` to be case-sensitive, so augment the selector to support both through an optional parameter.
The code currently supports both the baked-in homepages, as well as fetching from `/$/api/content/v1/get`.
As the latter is currently a circular round-trip, just use the baked-in version until the service is moved elsewhere. Loading performance will be better.
Potentially closes 1319
## Note
- `persistor.purge` returns a promise, we should wait for the resolved results before reloading, instead of just adding an arbitrary `setTimeout`.
- Perhaps the `setTimeout` method is to ignore a super-long or hanged purge (not sure), but it wouldn't be right since it would end up in a limbo state.
- Log errors to get clues for future.
- Reduced code by moving the reload to `finally`.
## Change
Due to "Own Comments" being implemented the same way as "Discussion Page", the comment lookup did not work for the abandoning case.
Just skip the local cleanup of `byId` -- it doesn't matter, as the GUI uses `selectCommentsForUri` to grab the per-channel comments, so it would still disappear in the GUI immediately.
## Issue
Ticket: 108
## Interesting
Uploads Page uses `claim_list`
Channel Page uses `claim_search
A second `claim_list` would instantly show updated results, but `claim_search` takes a long while to reflect the deletion.
## Change
This PR only handles the `claim_list` (Uploads Page).
The `claim_search` version would be troublesome to handle. If a refresh would bring back the results (due to the `claim_search` slow update behavior), we would need to store the deleted claim persistently. Then ...
- Store where? rehydrate?
- When to clear?
- How to make each `ClaimList` filter this out without polluting the code? etc.
...will be messy. Feels easier to "fix" at the `claim_search` level to make it update faster (maybe a command to clear cache at server or something).
Ticket: 1256
For `notify`, "file is currently locked" and "no such file or directory" is indication that the previous "failed" SDK call actually worked. Tell the user to check the transactions.
This is the band aid until odysee-api/401 is addressed.
## Issue
Accumulation of so many strings for a "successful tip" that it is getting annoying to re-translate them for minor changes.
## Change
- Consolidate and re-use strings.
- Make the tip string currency-agnostic so we don't have to touch it again. The amount and currency is moved to the `subMessage`.
- Fix spacing being incorrectly enforced through i18n (it should be done via code/css).
Ticket: 1189
## Issue
There is a bug in a batched `resolve` that returns jumbled data. Temporarily switch to `claim_search` until that is fixed.
## Notable differences:
- `resolve` will tell us directly that a claim has been removed or filtered, so redux will mark the ID as such. `claim_search` simply returns nothing, so we will still end up with an extra `resolve` for these items when the component tries to display it.
- The new function currently does not handle Collections (i.e. resolving individual items in the Collection) and Reposts. Given that this is temporarily, I'd like to leave `doClaimSearch` as is, instead of trying to replicate what's in `doResolveUris`.
## Notes:
Since we don't care if the resolve fails (and we weren't doing anything in the `catch` anyways), use `finally` instead.
- Remove the need for an object key. Just use an array. It gets hard to maintain as more messages come in, and so far there is no need to access the configs directly.
- Just focus on handling the message and not creating a toast object. Leave the latter to the client.
- Remove the odd resolve call that's lingering in `App`. Feels out of place, plus we don't need it now since the "active subs" claim search would contain resolved results.
- Don't search for active subs if Following count is zero.
## Issue
- Errors like "unexpected token at..." and "failed to fetch" are being logged, rather than the actual call error.
- Publish SDK timeout is sometimes successful, but the user doesn't know, so they re-created the same claim
## Change
- Attempt to make the logs more helpful, and the error message more meaningful to the user.
- Starting off with 'publish' and tip SDK call as an experiment. Can add others if this method works.
Was trying to save 1 state by assuming the homepage will be in a busy state and the ad-detection code will finish first. But this is not true for those with a small Following count.
* Factor out lighthouse-result processing code for FYP re-use.
The FYP results will be in the same format as LH.
* Recsys: add ability to pass in specific uuid to use
For FYP, we want to pass the UUID as a param when searching for recommendations. The search comes before the recsys entry creation, so we need to generate the UUID first when searching, and then tell recsys to use that specific ID.
* Redux: fetch and store FYP
Note that the gid cannot be used as "hash" for the uri list -- it doesn't necessarily change when the list changes, so we can't use it to optimize redux. For now, just always update/render when re-fetched.
* UI for FYP
* Mark rendered FYPs
* Pass the FYP ID down the same way as Collection ID
Not ideal, but at least it's in the same pattern as existing code for now. The whole prop-drilling problem with the claim components will be fixed together later.
* Include 'gid' and 'uuid' in recommendation search
* Allow users to mark recommendations that they dislike
* Pass auth-token to all FYP requests + remove beacon use
beacons are unreliable and often blocked
* Only show FYP for members
* FYP readme page
* small fixes
* fyp
Co-authored-by: Thomas Zarebczan <thomas.zarebczan@gmail.com>
Ticket: 1079 Support geoblocking channels/videos
## Changes
- Replaced the .env version with iapi version.
- Includes 'videos' blocking and custom messages.
* Add last used collection to right click menu.
* Small fix for last used collection when it's a published collection.
* Update last used collection when a collection is pending or published.
* Small refactor to get the last used collection.
We need to store the resolved claims first before marking COLLECTION_ITEMS_RESOLVE_COMPLETED, otherwise the GUI still sees undefined claims.
798 just accidentally flipped the order due to refactoring, I believe. It's now back to original.
## Cause
When 'selectRecommendedContentForUri' is filtering results against the blocklist, it relies on claim data which could be unresolved yet. It filters to empty results for this scenario, hence the flashing message.
## Change
Just return raw results when claims are not resolved yet, so that the GUI knows it needs to show the placeholders while resolving. After resolving, it'll go through the blocklist filtering again.
This is a follow-up for #1016
Still doesn't affect anything at the moment, but this reduce the amount of work needed when we yank out the homepages from our bundle later.
