From IndieWeb
(Redirected from indiearchive)

IndieArchive is a project to collaboratively grow an archival copy of pages replied to (possibly also mentioned) in indie web posts. was started as a possible shared repository of linked-to content. Presently it consists of a spec only, though several sites keep their own archive based on the concept.

URL design

Like, but without dumb/meaningless redundancy, and dates/times compressed with NewBase60 to make the URLs even shorter. example:

i-a design:

  • 3_C = 2005-03-10 converted to NewBase60 epochdays
    • e.g. using CASSIS ymd_to_sd() function
  • 3Kj = 03:19:44 = 11984 seconds in NewBase60
    • e.g. using CASSIS num_to_sxg() function


  • no "web." nor "/web" - we know this is a web archive, duh.
  • no "http:" - it's the default - optimize for the majority.
    • Other schemes may be explicit, but omit the "//" (per [1]), e.g. "https:", "ftp:" as shown in the example
  • Trailing slash is required on root-level domains

IndieWeb Examples

Aaron Parecki

Aaron Parecki kept an archive of external pages (both linked to and for webmentions received) from 2013-09-08 to 2016-01-31. As of 2016, the archive was 1.9gb in size. The archive contained only the HTML of the page, and did not include linked resources (images/css/js) of the external page. The primary purpose of the archive was to be able to parse the microformats in the page to render comments and reply context.

Barnaby Walters

Barnaby Walters is keeping an archive since ????-??-?? of external linked to pages, and has published a php library for maintaining and querying an archive:


The files in this archive should be stored on disk in an easily inspectable manner. However, URLs do not map cleanly to the filesystem. These four URLs are unique web pages but would result in conflicts on a filesystem:


Additionally, we need to be able to store http and https URLs separately, as well as non-standard ports:


As such, we need a simple set of rules to map URLs to filesystem locations, so that we can be sure there are no conflicts when saving on the filesystem.

Given a URL, the location on disk will be:

  • In a folder called "http" or "https" depending on the scheme
  • A folder for the domain name, all characters converted to lowercase
  • Split the domain, port and path on slashes (except a trailing slash), and create a folder for each except the last component
    • URL escape all characters not in a-z 0-9 . - _ (e.g. E becomes %45)
    • Append a : to the name of the folder (disambiguates between files and folders with the same name, and also looks nice since it appears as / on OSX in Finder)
  • Create a file for the last component
    • URL escape all characters not in a-z 0-9 . - _ (including the trailing slash if any)

For example, see below for the resulting paths for the examples above.

  • -> "http/"
  • -> "http/"
  • -> "http/"
  • -> "http/"
  • -> "http/"
  • -> https/"
  • -> "http/"
  • -> "http/"
  • -> "http/"

Note that without appending ":" to folders, it would be impossible to store all these files on disk since "" would have to be both a file and a folder.

This results in a folder structure that looks like the following (OSX renders ":" as "/")


Ideally I'd like to be able to store multiple versions of external files I encounter, for example when someone updates a post, I may want to be able to see the previous version of it so I know what has changed.

To accomplish this, it is easy to extend the above archive scheme to include versioning support by appending a version number to the end of the filename. A human-readable way to handle this it to use the current timestamp to indicate the version. For example:

  • http/
  • http/
  • http/

It is possible to find the latest version of a particular URL by sorting the files that match the base URL path alphabetically and using the last entry.


A conversation between aaronpk and tantek on IRC about link rot and an OSCON 2011 session that was about a WordPress plugin that automatically made personal archives of pages you linked to.[2]

So what would be a good way to make a backup of every page you link to from your site?

We could do it collaboratively for public pages, automatically in our publishing systems of pages we link to.


  • repair link rot. when things you've linked to go down or away, (semi-)automatically, redirect those links to your archives
  • shared reply/link contexts. see below thoughts from IRC.


On the design / usage of IndieArchive - notes from IRC:


  • Itch: "holy cow, I've already downloaded 63 megs of HTML from external pages in my reply context code"[3]
  • Scratch: "maybe we should just shove archive HTML pages into static github pages"[4]


  • keep snapshots by datetime index… we won't collide
  • we place no copyright claims and state it is purely for library/archive purposes only, and anyone is welcome to clone (not modify) and keep the same terms
  • once we get it going, we could even likely talk into maintaining a mirror of it - and it's git, so it's easy for anyone to mirror!
  • performance shouldn't be an issue either - as rarely should we need to reparse the HTML
  • still locally store parsed JSON bits for speed of retrieval / display on my server, but then keep a URL to the copy of the HTML on i-a.

Web of Social Trust:

  • we build up the write-access contributors through social web of trust from the initial seed of indieweb camp attendees that have their own indieweb sites setup - whatever they used to sign-into

Aside: IndieWeb vs. FSW methodologies and coming up with this

  • this is the kind of stuff that you end up figuring out / building when you scratch your own itches and follow them to their logical conclusions
  • I don't think anyone in FSW circles came up with the idea of a distributed collaborative mirroring of posts that they replied to! (despite it being a really simple idea)
  • everyone was so focused on building a full stack that did everything
  • instead of distributed systems that cooperatively grew

Aside: Unexpected uses

  • wouldn't surprise me if someone asks for a feed of all the URLs, as they happen, get linked to


The Internet Archive ( is a logical organization to coordinate with for indiearchiving.

2013-146 Brewster Kahle said to email him re: how can the Internet Archive help with IndieArchive.

on 2016-01-28 The Berkman Center released a plugin called Amber that caches linked pages using and


Some blue-sky speculation on the idea of "memory" in the web - both for remembering (archiving) as well as forgetting.

A site that notices certain off-site links are frequently clicked could begin archiving the target page locally - it retains a "memory" of the target content, and so that content has a greater chance of remaining over time. Conversely, a site that feel a need to do so, could choose to remove un-clicked links over time - it "forgets" about this portion of the web.

This could also be done directly by clients browsing sites too. Frequently visited sites are archived locally. Infrequently visited sites are just not archived ("forgotten").

See Also