Here we ask how a page is forked from one site to another?
# Current implementation
Wiki page-json is duplicated (copied) from one domain (wiki folder) to another. A page contains a record of it's entire history within the page-json schema, therefore preserving context and history of context.
# Long now structural considerations
How can we represent the current functionality, with the considerations of the permanent web? how do we ensure the integrity of an authors writing across millennia, as it moves and flows across the federation?
The strategy I personally wish to adopt here is to engage people who are thinking deeply about this question in conversation. Out of these conversations accompanied by software experiments we can begin to elaborate detailed answers.
# Experiments so far
So far I've not had time to go really deep into how this would work, but after a few iterations of experiments and writing, I see no obvious blocks to pursuing a permanent web strategy. i will try to outline this here as we move forwards.
DTB: my thinking on this hasn't changed for a while. I see pages as ordered lists of page-items as per json-schema.
However this can be impemented in a way that is congruent with the permanent web, with each item being represented / representable as an immutable content-addressable file.
The genetic / allele-style relationships between these items being represented by trees of such ids in which the family relationship is recorded.
Forking would copy the ordered list of item ids, to a new context list. Modifying an item would create a new allele of the item, and this would be recorded in the appropriate tree of allelic relationships.
At the same time a new list of items would be created for the modified page, in that particular context (name space = domain)
There is a tension between fluidity and immutability. However git, and genetics before that, have shown us that a great deal of fluidity can be derived from evolving branches of pointers to more substantial but fixed content.
A fluid evolving federation may struggle to implement a real-time robust permanence using existing technology. However we can imagine taking regular snapshots.
It is not fully clear how such snapshots would relate to the current more fluid implementation, but we imagine recording a link to the snapshot in the journal, and extracting and indexing other information like idem-id and allelic relationships at that time.
# See also