It's still under development, and there are some asterisks to apply to the data sources, but I find the directional data you can get out of the new HTTP Archive Tech Report page helpful:
-
It's still under development, and there are some asterisks to apply to the data sources, but I find the directional data you can get out of the new HTTP Archive Tech Report page helpful:
-
This, e.g., is instructive:
-
@slightlyoff Gee, wow, 'only' 2MB. So svelte
-
@fugueish In fairness, that's total page weight, not JS transfer...which is...a different number -- though not a better one per se:
-
@fugueish But cool to see that all the work to get to "React Server Components" (which is what "Next.js App Router" means) has shaved off...
[ checks notes ]
...5% of the JS.
-
@fugueish (for the avoidance of doubt, that's a joke. Next.JS w/ RSC is in the noise, statistically speaking, for scale, so it shouldn't be discussed as a serious factor. Next vs Lit, though, seems to be in the same order of magnitude and a more reasonable comp)
-
@slightlyoff I know. My rule, though, is: "The ratio of human-meaningful information to bytes transferred must be high; or, for fuck's sake, at least 1.0." 1 MiB of text is about 500 pages of ASCII, and it IMO actually is fair to say an image is worth 1,000 natural language words. So, for 1.02 MiB, I want a gd novel and maybe some `onclick` handlers for flair. Or an illustrated novella. 1,000-word news articles with a photo... not 3 MiB.
This post brought to you by Mastodon's web front-end
-
@fugueish I hesitate to point out that it's transfer, so that's 1MB of *compressed* text.
/me ducks
-
@slightlyoff I was trying to give them a mulligan...
-
@fugueish After a decade of pulling punches, I've decided that most of the methodological graces are mistakes. They've mostly served to insulate the shysters from legitimate scorn.
