hacker-news-pwas
hacker-news-pwas copied to clipboard
Standardize specifications
Some discussion has been made to standardize specifications between the different implementations - let's use this ticket to discuss how we can/should do this.
cc @anubhav7495 as he's been helping out with extensive reviews recently. cc @kristoferbaxter in case he has opinions here too :)
Some of the discrepancies that I've observed in PRs we've received that would be useful to document better for the specification:
- Mobile-only layout (vs. a responsive layout that has suitable margins for wider screens/desktop)
- Folks forgetting to have views for child pages (e.g comments) for all top-level navigation items.
- Forgetting to check at least the application shell is caching correctly. Should we be standardizing on data being cached too? Most folks don't seem to have had issues accomplishing this step.
- Standardizing on a routing scheme? We usually mention this nowadays citing apps like PreactHN, but don't call it out much in the current spec.
+1 to all of these
The data caching step isn't currently supported by a few implementations, so we could either:
- Ask the authors to update with a reasonable timeframe 1a. Should implementations be dropped if they are not updated? 1b. If not, what should be done?
- Implement PRs for existing implementations
This sets a precedent for future additions to the specifications.
:+1: from my side too.
Also, some older apps list 20 stories per page instead of 30. We should also open PRs for those implementations.
Examples: Svelte HN Vue HN Viper-News Nuxt HN
How do y'all feel about standardising the data fetching, CSS, routing(?) etc so that the examples are more purely focused on the UI implementation side of things? Similar to todomvc-common.
what about sites hosted on different platforms that will result in different response times? will be ideal to have a single host provider to avoid any response discrepancy between the different providers
also all demos should be single page website, some demos are some aren't, but adding this behavior can add some extra kb to the request, having a direct impact on the speed.
ps: would be nice to add as a metric the js size on the initial page load.