Skeleton of IPLD v1 spec #12
Conversation
|
LGTM |
|
I have just added abstract and intro, @RichardLitt can you have a look? |
|
|
||
| > This specification defines a data model and a naming scheme for linking data with cryptographic hashes. | ||
| > | ||
| > InterPlanetary Linked Data (IPLD) is an information space of inter-linked data, where content addresses and links are expressed using the content's cryptographic hash. IPLD is designed to universally address and link to any data so that the data does not require a central naming authority, the integrity of the data can be verified, and untrusted parties can help distribute the data. This specification describes a data model for structured data and a naming scheme to point to data and subsets of the data. These design goals make it different from earlier data models such as JSON or RDF, and naming schemes such as NI [[RFC6920]](https://tools.ietf.org/html/rfc6920), or Magnet URI. |
There was a problem hiding this comment.
an information space of inter-linked data
Seems very odd phrasing to me.
content addresses and links are expressed using the content's cryptographic hash
Which is expressed as a hash? The addresses or the links? Are they the same? Why say links then?
There was a problem hiding this comment.
@RichardLitt that is true, I need to check with Juan on how we want to clearly define IPLD. Technically, information space is a valid wording (the web, for example, is an information space; an information space is made of linked documents) - maybe we can just replace that with
IPLD is a web of data, where data is .. and so on
What do you think?
There was a problem hiding this comment.
No, information space works.
|
Gave it a quick pass and added a comment. Let me know when more is written! Seems good so far, if dense. |
ipld-future/README.md
Outdated
| 1. **Data integrity**: URLs give no guarantees that the data we receive hasn't been compromised, the IPLD naming system ensures that no one can lie about the data they are serving | ||
| 2. **Distributed naming**: only the owner of a domain name can serve you the data behind a URL; in IPLD any computer - trusted and untrusted - that has the data can help and participate in distributing it | ||
| 3. **Immutable Content**: The content behind URLs can change or disappear, making our links broken or not pointing anymore to what we were expecting. IPLD links cannot mutate. | ||
| 1. **Data integrity**: URLs give no guarantees that the data we receive hasn't been compromised. The IPLD naming system ensures that no one can lie about the data they are serving. |
There was a problem hiding this comment.
thank you!
Previous work especially the RFC on "NI" was referred to as "naming things with hashes".
We can choose either wording and stick with one, I like addressing since naming systems are often confused with DNS replacements
There was a problem hiding this comment.
Exactly, that is why I prefer addressing more
| @@ -0,0 +1,179 @@ | |||
| # IPLD Spec v1 | |||
|
|
|||
| Editor: Nicola Greco, MIT | |||
There was a problem hiding this comment.
If there is an editor field, might be good to have a list of reviewers as well?
There was a problem hiding this comment.
yes, the spec will contain all the names of the people that have worked on
I have drafted a possible skeleton for the IPLD v1 spec.
The key point of this draft is defining the different topics and separating IPLD into IPLD pointers, IPLD data model and IPLD representations
We can find better keywords on our way