## Ticket
1526: strange thumbnail size requested on mobile layout (pc only?)
## General Problem
It was trying to fetch based on the exact size of the video container, which would satisfy Core Vitals (in an overkill way), but would bring several issues:
- server-side caching would not work since everyone's window size is different in a responsive layout design.
- the additional 200ms wait for container size to settle down is not good (hardcoded wait time).
- the code did not account for device-pixel-ratio, so it's quite a futile effort.
Aside: In the past, we used to take the same image url as the tiles, so the video poster would appear immediately from due to browser cache, but the quality is bad because the tile requested a much smaller size.
The embed wrapper was not going through the CDN either as a null `containerRef` was passed in.
## Change
Removed the container-size check and just request for 1280x720. Reasons for this size:
- On average, that would be the ballpark of the final calculated value anyway for the average screen (+DPR consideration).
- That seems to be the current suggested thumbnail size in most recommendations.
- Our YT Sync is grabbing a much smaller size anyway.
## Ticket
1583: "add incontent ads to category/channel pages (break up every X claims?), or add back bottom ad in those areas."
## Behavioral changes
- Hide when in homepage (as currently we have ads between categories).
- Fix the light theme (it was transparent).
## Code changes
- While an Effect is the 'right' choice given that there is no jsx to mount, the need to prop-drill from the parent was getting a bit annoying, so converted to a Component instead.
- Remove the delay for Core Vitals avoidance for now -- seems to make the sticky less likely to serve an ad.
- Now that the membership state is correctly populated for incognito (see 9d830615), there is no more need to check for `isAuthenticated`.
Using that global variable was a bad idea. Stick to redux as the source of truth.
The flag is not listed for rehydration, so it will always start off as `undefined`, which is what we need anyways.
`undefined` indicates we haven't tested for ad-blockers, so we'll run the fetch and update the store with a true|false value.
Previously, we hide when authenticated.
We still need the `isAuthenticated` flag, because `hasPremiumPlus` stays at the unfetched (`undefined`) state for Incognito.
It's also possible to move the use of `useClaimListInfiniteScroll` into `claimList` itself, but opted to keep them separate for now so that `claimList` doesn't end up a bloated file like `claimListDiscover`.
In the first attempt (PR 1497, 4d44d81f), the `generateStreamUrl` was tweaked to fetch a `Lbry.get`, which returns a promise. On top of a few async issues on its own, we also can't just simply `await` here since render functions need to be pure.
Fortunately, there is already redux structure for `get`, and Sean happened to already be passing all the required stuff here (sight beyond sight!), so it's just a matter of calling it.
Again, render functions need to be pure, so the fetch has be in an effect, so the component might take several renders to get the image painted. But oddly this seems to avoid the occurrence of showing the fallback spaceman image (just an observation, no proof of correlation).
It was meant to be general-purpose in the first-place, but ended being tied to ads due to time constraint and also bad documentation.
No functional change.
## Issue
- Due to the timer used, the script could be loaded after the path-checking useEffect ran.
- Stale closure is also an issue.
## Approach
Create refs for the props, and listen to `onload` as well to resolve the visibility state.
pinnedUrls = resolve
pinnedClaimIds = claim_search
Make `pinnedClaimIds` take precedence for now, since the homepage is defining both to support Android.
For some reason, I previously made `pinnedUrls` the priority, thinking this should be the "correct" one to use when the SDK is fixed.
Expanded homepage pins to support 2 types of input (if both are passed in, pinnedUrls take precedence):
(1) pinnedUrls --> uses doResolveUris (resolve)
(2) pinnedClaimIds --> uses doResolveClaimIds (claim_search)
Instead of injecting the pinned URLs directly, we inject from `resolvedPinUris`, which will be blank until the uris are resolved, thus preventing it from being resolved individually from the tiles.
There's additional complexity with the `claim_search` method, as the rest of the code deals with uris instead of IDs. Fortunately, it's all contained with `useResolvePins`, so removal would be easier when the batch `resolve` issue is fixed.
- Only add for unauthenticated users for now. We can change that to look at premium later without having to touch `app/view.jsx`.
- Only show the ad after knowing the ad is filled; this prevents the invisible container from blocking stuff while waiting, or even worse, when not filled.
- Instead of 2 ways to display ads (DOM injection + React method) and having both of them clash, just do it the predictable React way.
- Augment the existing React version to support tile layout + ability to place in last visible slot.
- Consolidate styling code to scss ... DOM manipulations were making it even harder to maintain.
- Removed the need to check for ad-blockers for now. It was being executed every time an ad is displayed, and now that we are displaying ads in more places, the gains doesn't justify the performance loss. Also, it wasn't being done for Recommended ads anyway, so the inconsistency probably means it's not needed in the first place.
Other known issues fixed:
- double ad injection when changing language via nag.
- additional "total-blocking-time" due to ads at startup removed.
- fixed ads not appearing in mobile homepage until navigated away and back to homepage.
- enable ads in channel page.
- support for both List and Tile layout.
* Bump react-draggable
Old version was giving out console errors for outdated react functions
* Refactor fileRenderFloating
* Merge fileRenderMobile into fileRenderFloating
* Fixes from review
* Attempt fix failed to view live
