Aral Balkan
(Re)introducing 404 → 307 (the evergreen web technique)¹
What if links never died? What if we never broke the Web? What if it didn’t involve any extra work?
It’s possible. And easy.
Just make your 404s into 307s.
¹ Previously, I was doing 404 to 302 but 302 redirects do not keep the request method so 307 is the correct redirect to use. I’d been meaning to update the site for ages and so I finally made some time to do so this week.
#web #dev #evergreenWeb #4042307
Aral Balkan
And, of course, Kitten now has native support for 404 → 307.
https://codeberg.org/kitten/app#404-307
(Currently for developers to play with via the Kitten.json file but soon to be exposed as a setting in the web-based settings interface.)
#Kitten #SmallWeb #4042307 #web #dev #evergreenWeb
Aral Balkan
@jkmcnk Depends on the content. We archive information to preserve knowledge. We owe a vast chunk of our knowledge today to the work of archivists through the ages. Today, just because we write on the Web and not on parchment, it doesn’t mean the importance of preserving our work is any less. And, unlike parchment, our thoughts in hypertext are intertwined with those of others. We don’t break URLs so as not to harm that fabric.
jaKa Močnik
@aral tbh, links and content dying and going away is a blessing, not a curse. not sure where this idea that everything on the internet should stay forever comes from. things change and things die. 🤷
Aral Balkan
@jkmcnk That, and I really hate broken links ;)
Aral Balkan
@ramsey That’s on them. We do the best we can :)
Ben Ramsey
@aral I recently did this and found that lots of older clients that used my RSS feed wouldn’t follow the 307 redirect because they hadn’t been updated in 10-15 years.
Aral Balkan
@ramsey Fair enough. Would be good to document that. (I’ll make myself a note to do so.)
Ben Ramsey
@aral I switched back to 302 because too many of them were failing to redirect.
Aral Balkan
@hananc Good feedback, thank you :)
חנן כהן • Hanan Cohen
@aral 404 is well known but 307 is much less so. I had to read deep in the site in order to learn what this code is. I suggest mentioning on the first page that 307 is temporary redirect.
Aral Balkan
@leitmedium If your previous site is a dynamic site then any POST, etc.., requests will fail. Especially if they are absolute URLs instead of relative.
If this isn’t a concern for you, then by all means use 302 for greater compatibility.
Caspar C. Mierau
@aral Interesting, thanks. What I did not get fully yet is the need to do a 307 instead of a 302 here. Besides a decreased client support and differences in the handling of HTTP method changes, I do not see a real difference. Can you tell me where I can find more reasoning for the 307 over 302 decision?
argv minus one
I'm responsible for a website that has *thousands* of commits to its repository. Publishing every last one of them and chaining 404s like that would be a great way to DDoS myself.
Aral Balkan
@argv_minus_one Sorry if it wasn’t clear, this isn’t about keeping every version of your site (every commit, etc.) available. It’s for when, say, you migrate your site from one platform to another or from a blog to something else, etc., and you don’t want URLs to break without having to host a static export, etc. It’s definitely not about making every version of your site history accessible (your URLs are likely not all changing between commits either).
Aral Balkan
@serge @argv_minus_one Yep, use whatever works for you. This is one way that works for me and will be easy for folks to implement on the Small Web without requiring too much tinkering.
I also use the archive to a directory method and Site.js – the precursor to Kitten – calls them archival cascades and supports them natively too (https://codeberg.org/site.js/app#user-content-the-archival-cascade)
Serhii
@argv_minus_one @aral I also don't see how maintaining a dynamic site would be feasible, both cost-wise and security-wise.
I think the most practical way would be to convert your website to a static one, if content is what you want to preserve. Virtually any web server setup supports serving static files from a public directory (or can be set up to do so), so you can just dump archived pages to a subdirectory and add a redirect to there. Since they're static files, a simple 301/302 would suffice.
That requires being in control of both the domain you're migrating from and your new server configuration. So yeah, this won't help if you're moving from a *.wix.com site or to somewhere where you cannot add redirect rules.
argv minus one
Well, the site I'm responsible for is fully under my control, and I can transfer it in its entirety to another host at will, but…
…if it wasn't, and an old version was hosted on Wix or something, wouldn't I then have to keep paying for that old platform indefinitely even though almost no one visits that old version?
０７６萌ＳＮＳ is a social network, courtesy of ０７６. It runs on GNU social, version 2.0.2-beta0, available under the GNU Affero General Public License.
All ０７６萌ＳＮＳ content and data are available under the Creative Commons Attribution 3.0 license.