Accept-Language is unreliable. Most browsers ship a default of en-US,en;q=0.9 regardless of the user’s actual preference, and most users never change it. Treating it as a preference signal is guessing.
Well, the alternative is to always land on the wrong language. I spent a few years in Switzerland and I was tired being sent to the German version of the website instead of the site respecting the language of my browser. For mobile users, Accept-Language is derived from the OS language. For desktop users, if you install Firefox using the French-localized installer, you'll get French as the default language. I don't know how it is for Chrome. It is not perfect, but it is still far better than using a random value.
I spent a few years in Switzerland and I was tired being sent to the German version of the website instead of the site respecting the language of my browser.
I also get that sometimes, with the preferred language in the OS and Browser set US-English and German as a fallback. My impression is, that this sites are doing it based on geo-IP lookup from where you are connecting and not based on the language settings in the browser.
It was shaped like a good idea. I've wanted a linter that would check a number of these points for quite some time. And it might not even be a terrible starting point for a human who wanted to check sources and curate their own list.
It's exhausting to have to dig to see if something's slop every time you find something that seems worth sharing. I might have shared this too, if I found it when I wasn't in the mood to go dig into its git repository and pick apart the sourcing and categorization for each point. The other comments about mismatches between headers and content are accurate, IMO, but I'd have missed them too, because there was enough here that seems useful before digging in.
The section headers can be a bit odd; I'm not sure why rel="canonical" is in the Foundations section and not the SEO section and I'm not sure why a web app manifest is in the Resilience section. And yeah, I think you need to take the optional/recommended tags with a pinch of salt and decide for yourself based on each site's context (as others have commented).
The section headers can be a bit odd; I'm not sure why rel="canonical" is in the Foundations section and not the SEO section and I'm not sure why a web app manifest is in the Resilience section. And yeah, I think you need to take the optional/recommended tags with a pinch of salt and decide for yourself based on each site's context (as others have commented).
Yeah, I'd agree. On second thought though, uh, the taxonomy seems a bit LLM-vibed...
Though I do have a few gripes with it.
For one, I'll skip llms.txt, and I should probably be stricter on URLs and such. I don't agree with stripping query parameters in the canonical though.
HTML is already a semantic markup language… with microdata, you can skip the JSON-LD stuff too. You actually lose a ton of meaning converting to Markdown (which if it must be Markdown, why doesn’t the file extension reflect that, & if it is plaintext, why call it Markdown?).
vbernat | 7 hours ago
Well, the alternative is to always land on the wrong language. I spent a few years in Switzerland and I was tired being sent to the German version of the website instead of the site respecting the language of my browser. For mobile users, Accept-Language is derived from the OS language. For desktop users, if you install Firefox using the French-localized installer, you'll get French as the default language. I don't know how it is for Chrome. It is not perfect, but it is still far better than using a random value.
fab23 | 5 hours ago
I also get that sometimes, with the preferred language in the OS and Browser set US-English and German as a fallback. My impression is, that this sites are doing it based on geo-IP lookup from where you are connecting and not based on the language settings in the browser.
blinry | 7 hours ago
Apparently pumped out using Claude in a single day. That means that it's not something I'm very interested in using or contributing to.
[OP] runxiyu | 3 hours ago
Should have spotted that earlier. I'll flag my own submission as spam I guess
Edit: well looks like I can't flag my own submissions
hoistbypetard | 2 hours ago
It was shaped like a good idea. I've wanted a linter that would check a number of these points for quite some time. And it might not even be a terrible starting point for a human who wanted to check sources and curate their own list.
It's exhausting to have to dig to see if something's slop every time you find something that seems worth sharing. I might have shared this too, if I found it when I wasn't in the mood to go dig into its git repository and pick apart the sourcing and categorization for each point. The other comments about mismatches between headers and content are accurate, IMO, but I'd have missed them too, because there was enough here that seems useful before digging in.
jarofgreen | 8 hours ago
The section headers can be a bit odd; I'm not sure why rel="canonical" is in the Foundations section and not the SEO section and I'm not sure why a web app manifest is in the Resilience section. And yeah, I think you need to take the optional/recommended tags with a pinch of salt and decide for yourself based on each site's context (as others have commented).
The SEO section could also make clearer robots.txt is not actually how you stop your page appearing in Google, it's the robots tag listed further down. And those 2 things need to work together - see Googles official page especially the "Combining robots.txt rules with indexing and serving rules" bit at the end. This is a very common misunderstanding so seems worth being very clear on.
But to be clear, the fact I'm nitpicking says that generally this is a very good list and there were a few things I didn't know. TIL! So thank you.
ps. The site encourages contributions so do pass on thoughts upstream too!
[OP] runxiyu | 7 hours ago
Yeah, I'd agree. On second thought though, uh, the taxonomy seems a bit LLM-vibed...
jarofgreen | 6 hours ago
Just checked and Claude is involved. I'll remember to double check everything.
[OP] runxiyu | 8 hours ago
Some of these lists are truly useful.
Though I do have a few gripes with it. For one, I'll skip
llms.txt, and I should probably be stricter on URLs and such. I don't agree with stripping query parameters in the canonical though.gerikson | 8 hours ago
I’m gonna ensure my sites actively avoid anything in the ” Agent Readines” section.
Not sure I’d put HSTS as ”required”, either.
[OP] runxiyu | 7 hours ago
I mean, stable URLs are nice.
gerikson | 6 hours ago
They just put that in there to mess with statements like mine 😄
Could just has well been in the SEO section.
toastal | 2 hours ago
HTML is already a semantic markup language… with microdata, you can skip the JSON-LD stuff too. You actually lose a ton of meaning converting to Markdown (which if it must be Markdown, why doesn’t the file extension reflect that, & if it is plaintext, why call it Markdown?).
olegkovalov | 7 hours ago
It would be awesome to paste a link and see what is points are valid.
antonmedv | 4 hours ago
A lot of obvious and not useful checks.
No new insights