## Issue
6542 Livestream listing and blocked list kills CLS score
This is basically a repeat of what was done on `claimTileDiscover` (homepage), but now on `claimListDiscover` (category pages).
This handles the unnecessary re-render when:
- the uri list temporarily being zero while waiting for new claim-search results.
- livestream claim-search invalidating the list.
## Changes
Store the last uri list and use that when we know that claim-search is still not done.
## Tests
1. Ran Lighthouse on category pages (force a livestream channel on it). CLS score must be green.
2. Block `claim_search` from Dev Tools and move to another category. The page should say "timed out" instead of "no results" (i.e. maintain existing behavior).
## Issue
6589
## Notes
`parseURI` will return `isChannel=false` due to the reposted channel URL not having the '@' symbol.
Since we already have the claim object, check `value_type` directly instead.
Pop up Menu Fixes and Improvements
Improve re-directs
Fix file deletion
Improve code, and more
Fix List Layout collection popup
Allow to edit list
Fix blocking when not logged or no channel
Fix Edit and Delete showing for not owned Lists
Fix actions breaking when not logged in
Fix List options not showing
Lint
Shorten logic
Lint
Recently changed to `<OptimizedImg>`, which was assuming the `src` never changes.
dejavu ... we also hit this issue when implementing the `IntersectionObserver` lazy-loading.
Objective:
- Get appropriately sized images to improve performance and Core Vitals score.
- Ensure images using "objectFit=cover" doesn't get stretched out if the source is large enough.
- Peg to 100px increments for better caching.
Notes:
- Skip images hosted in '/public'. If we really want to optimize it, then we'll need to provide the full path in the code, otherwise CDN lookup will fail.
If there are no API errors but no reactions returned, consider the requested IDs as "done" and stop requesting again. When a channel is being confirmed, Commentron doesn't return the reaction object.
Also added extra guard in case Commentron does return an object in the future, but an empty one.
- use `selectActiveChannelClaim` as that takes the current channel list into account (i.e. correct state when all channels are deleted).
- `selectActiveChannelId` should probably be removed or not exposed through a selector, as it is not updated when channel list changes?
## Issue
6068
## Change
Lock all images to fit a 16:9 container. We have implemented 'ZoomableImg', so no reason in trying to display in full size.
This reduces CLS from 0.4xx to 0.01x.
## Flaws
CLS could probably be zero if not for the spinner shifting things slightly. Also, mobile CLS is 0.07.
The troublesome part in this PR is that FileRenderInitiator, FileRender and their subcomponents are broken apart and it's hard to synchronize their visibility and size. There are time gaps where none of them are visible, etc.
This PR only tackles the major part (most bang for buck), which is the elimination of variable height of the rendered image.
## Aside
I think `claimIsMine` is unused, so don't waste time requesting it.
show visible card and add remove card button
show your transactions even if you dont have a card
fix presentational issues
show your transactions even if you dont have a card
fix presentational issues
add link to channel section
update yarn
show donation location
add remove card modal still needs completion and also changed how stripe is used on settings stripe card page
add confirm remove card modal to router
move bank account stuff to settings page
move account functionality to settings page
continuing to move account transactions to settings
list transactions for creator
updating copy
touchup tip error
do a better job autofocusing
bugfix
show an error on the card page if api returns 500
building out frontend for comment tip
display dollar sign if its a fiat tip
more frontend work
more frontend work
more frontend bug fixes
working with hardcoded payment intent id
working but with one bug
bugfixed
add toast if payment fails
add add card button
cant get claim id but otherwise done
more frontend work
call is working
show fiat for livestream comments
add is fiat on comments
round and show values properly
dont allow review if tiperror
copy displaying properly
disable buttons conditionally properly
remove card button working
remove card working with a workaround by refreshing page
bugfix
send toast when tip on comment
jeremy frontend changes
only show cart on lbc
In the event that Commentron returned an empty object, we'll end up re-requesting the same IDs again. Haven't seens this happen before, but since we don't know what's causing the spike, we'll just consider failures and "fetched" to stop the loop.
User can always click Refresh to repopulate the values.
Not sure if this will ever occur, but technically possible from the code point of view. Try adding this to see if it stops the spikes in `reaction.list` calls.
Someone mentioned about odysee.com showing "tip" related text when supporting own claim. Now it should default to "Boost" related text for own claims.
L167: Not edited but not sure if "claimIsMine" is needed there any more.
add watch later hover action
replace watch later popup item for favorites
lint
styling for watch_later overlay
Add label
Use just claim, add requiresAuth
Add list icon
Tone down text
Turn WL Hover Button into component
Change WL hover icons
small revert
Keep watch later in the menu
## Issue
- `Comment.replies` currently represent all replies, while `comment.List` returns a filtered version, so the actual replies could be less.
- The actual replies is represented by `total_filtered_items`, but we only get that after making a fetch. So, users could click "Show more" but get nothing.
## Fix
- Stop showing "Show more" based on `total_filtered_items`.
- If there is a balance, display 1 dummy comment to represent all blocked replies. This handles the case of "Show more" being displayed but ended up with 0 replies if all replies were blocked.
## Future
Note that `Comment.replies` might be changed to represented filtered comments in the near future (refer to Beamer), so the GUI is made such that the dummy just won't appear when that change happens.
Seeing the spinner too much can be annoying.
- This approach works, but currently, when the list is very long, something is taking up resources and the handler couldn't be processed, so the effect is lost (still seeing the spinner). See 6473.
- Since we are now prefetching, bumped the debounceMs a bit.
## Issue
6414 <Don't use "exact match" setting outside of main search - cuases no results in related>
`doSearch` was used in Wunderbar and Recommended, but it was adding in Wunderbar options all the time.
## Changes
- Fix `doSearch` to not use hardcoded options (take from arguments).
- Refactors the searchOption-creation code so that we don't duplicate the logic in both .js and .jsx.
## Root-cause
`https://github.com/videojs/video.js/pull/7098`
videojs was recently upgraded.
## Change
Override videojs default theme of hiding the time control in smaller layouts.
## Issue
#6332 <Thumbnail lazy-loader is too slow || Use browser-level lazy-loading>
## Change
Switch from "threshold-based check" to "viewport distance comparison" using the `rootMargin` parameter. The root is the viewport.
This change makes it closer to the native `loading="lazy"` behavior, where it starts to load when approaching the viewport. Chrome I believe uses 3000px distance -- I think 500px is a good compromise for now. Can adjust further.
## Future
- We are currently creating N instances of IntersectionObserver.
- https://developers.google.com/web/updates/2016/04/intersectionobserver
- "If you need to observe multiple elements, it is both possible and advised to observe multiple elements using the same IntersectionObserver instance by calling observe() multiple times."
This would probably need a refactor to make ClaimList (or something higher) own the IntersectionObserver.
It was waiting too near the viewport to load on certain zoom levels. This will maintain the same physical distance for the "approaching viewport" buffer regardless of zoom levels.
## Changes
- Change to canonical_url (with ':').
- The old format of 'claimName/claimId' needs to be supported still, since we shipped with it.
## Notes
- It would be nice to use regex instead of 2 separate paths, but I couldn't figure out how to make the koa variables work.
We currently handle major errors through ErrorBoundary. If the error is network related, ErrorBoundary ends up blank if it's components are lazy-loaded.
#codesplit
It won't work with the hardcoded value.
Still waiting for commentron (error code + data) enhancement. It's pretty weird having to parse the number, then add it back in.
## Issue
Sub item in <6119 Creator Settings: beyond "Muted Words">
## Changes
- Removed the existing 5s min gap on livestreams.
- Enabled the 'slow mode min gap' in Creator Settings
This change now affects both comments in claims and livestream comments.
## Issue
The loading circle wasn't showing up, causing confusion.
Splitting CSS doesn't seem trivial as there seems to be a huge dependency on the load order. Pretty much similar to what this person is facing https://lihautan.com/css-code-splitting/#the-problem
## Change
This reverts videojs-specific changes from 4d638bcf.
## Issue
When seeking on a video, RecommendedContent gets re-rendered, which causes the 20 ClaimPreviews to be re-rendered, several times. ClaimPreview has lots of children, plus it's constantly checking against the blacklist, among other things.
## Fix
The culprit seems to be the `recommendedContent` array -- the address may be different, but the contents are the same.
Do a deep compare instead.
- [x] (6332) The IntersectionObserver method of lazy-loading loads cached images visibly late on slower devices. Previously, it was also showing the "broken image" icon briefly, which we mended by placing a dummy transparent image as the initial src.
- Reverted that ugly transparent image fix.
- Use the browser's built-in `loading="lazy"` instead. Sorry, Safari.
- [x] Size-optimization did not take "device pixel ratio" into account.
- When resizing an image through the CDN, we can't just take the dimensions of the tag in pixels directly -- we need to take zooming into account, otherwise the image ends up blurry.
- Previously, we quickly disabled optimization for the channel avatar in the Channel Page because of this. Now that we know the root-cause, the change was reverted and we now go through the CDN with appropriate sizes. This also improves our Web Vital scores.
- [x] Size-optimization wasn't really implemented for all ChannelThumbnail instances.
- The CDN-optimized size was hardcoded to the largest instance, so small images like sidebar thumbnails are still loading images that are unnecessarily larger.
- There's a little-bit of hardcoding of values from CSS here, but I think it's a ok compromise (not something we change often). It also doesn't need to be exact -- the "device pixel ratio" calculate will ensure it's slightly larger than what we need.
- [x] Set `width` and `height` of `<img>` to improve CLS.
- Addresses Ligthhouse complaints, although technically the shifting was addressed at the `ClaimPreviewTile` level (sub-container dimensions are well defined).
- Notes: the values don't need to be the final CSS-adjusted sizes. It just needs to be in the right aspect ratio to help the browser pre-allocate space to avoid shifts.
- [x] Add option to disable lazy-load Channel Thumbnails
- The guidelines mentioned that items that are already in the viewport should not enable `loading="lazy"`.
- We have a few areas where it doesn't make sense to lazy-load (e.g. thumbnail in Header, channel selector dropdown, publish preview, etc.).
## Issue
The scroll listener never unregisters, and is always registering itself on every scroll. I believe it was done that way to also handle the case of "element is already in viewport when mounted".
## Change
Tried to separate both "element is already in viewport when mounted" and "element scrolled into viewport" into different effects.
The timeout value used is a bit arbitrary, but is needed because the initial size is (0, 0), and to debounce any layout shifts. Reasoning: If an element is explicitly placed under this wrapper, the additional delay is acceptable since it's meant to be lazy-loaded anyway.
about to test something
generate programatically
beginning of the frontend
stripe integration page seems to be working
add user
put functionality behind conditional tag
connect frontend working well
adding environment variables to save success and failure url
bugfix
bugfix
final clean up
adding credit card page
seems to be coming along
calls successfully coming from the frontend
fixing up frontend
cleaning up
frontend coming along
client secret working
basic frontend in place
adding tip page
adding more to the tip frontend
frontend almost done
tabs coming along
one last thing to do for frontend
adding explainer text as custom function
putting finishing touches on tabs
support tabs working well
disable fiat toggle when card not connected
fix frontend gui bug
bugfix and pull out label function
fix symbol for tip gui
modal when card is not yet saved
fix fiat disabled bug
knowing whether card is added programatically
sending tip with frontend
tip functionality working
show unpaid balance
add frontend for card add section
update frontend
update frontend
bugfix
change to use react instead of css
update how stripe is instantiated
fix bug
use customer setup
coming along
working but needs optimization
persist if card is saved
adding anonymous tip functionality
fix nan bug
build stripe endpoints programatically
show for all users for time being
allow the stripe key to automatically switch to live environment
bugfix
bugfix
fix jslint
fix channel page support button
better docs
show customer transactions on frontend
basic table in place
various page updates per jeremys notes
showing card details
nicer tip history table
add better prompt to add card on file viewer page
some linting
time
put connect account behind fiat enabled
no persist fiat mode
wallet calls
tip stuff
## Issue
3779 RSS feed for channels
## Initial implementation details
- RSS only (not atom)
- Grabs latest 10 entries (Beamer have concerns)
## Credit
Referenced the community version mentioned in 3779
- The `rateChange` event now logs the updated speed,
not just the time at which it occurred.
- The `scrub` now (more) accurately logs the position
it came from before the destination.
- The recsys events get consolidated for logical consistency.
## Issue
With autoplay off, clicking on another video from the Recommended section didn't update the video poster.
## Fix
Made a mistake assuming that the whole component will be re-mounted, so was only setting the thumbnail once.
Even with the caching changes, a 150kB thumbnail still takes 1-2s to fetch. This impacts the score.
## Change
(1) Start with a large-enough placeholder image (has to be larger than the final image -- inflating doesn't count), then delay just enough for scoring, then switch to the real thumbnail.
(2) Since we are now doing post-mount stuff, we have the exact dimensions to optimize the claim thumbnail. This reduces the typically-several-MBs thumbnail to kBs.
These are chunks that will be requested immediately after ui.js when opening the homepage, so consolidate them into 1 chunk to reduce network overhead.
More work can/should be done to reduce the secondary chunk.
Make immediate GUI feedback to convey the current status, which can be the following:
- typing
- waiting lighthouse results
- waiting claim resolve
- no results or failed.
## Issue
6301: don't use CDN for GIF thumbnails
## Note
Just a minor mistake in the recent thumbnail refactoring; left out the handling when `allowGif` is true.
While Lighthouse allows looser searching like "Tom from LBRY", it doesn't show the expected results when direct channel name with partial ID is entered to disambiguate.
