add terminology doc

This commit is contained in:
Brad Fitzpatrick 2010-06-20 18:18:43 -07:00
parent 65bdc67cec
commit d9e9c6c89b
1 changed files with 70 additions and 0 deletions

70
doc/terminology.txt Normal file
View File

@ -0,0 +1,70 @@
We use the following terms in our documentation and code:
Blob: 0 or more bytes, with no extra metadata
BlobRef: a reference to a blob, consisting of a cryptographic hash
function name and that hash function's digest of the blob's bytes,
in hex. Examples of valid blobrefs include:
sha1-f1d2d2f924e986ac86fdf7b36c94bcdf32beec15
md5-d3b07384d113edec49eaa6238ad5ff00
sha256-b5bb9d8014a0f9b1d61e21e796d78dccdf1352f23cd32812f4850b878ae4944c
Concatenating the two together with a hyphen is the common
representation, with both parts in all lower case.
Object / Camli Object: a data structure consisting of keys/values
where values are string literals, numeric literals, other objects,
or arrays of values. In Camli, these are serialized as JSON
everywhere.
Object Blob / Camli Blob: the JSON serialation of a Camli Object.
This is just a normal JSON serialization (with any JSON
serialization library that's available), with the extra requirement
that the serialization must start with the 16 bytes:
{"camliVersion":
With no extra whitespace, no single quotes instead of doublequotes,
etc. Those 16 bytes are the magic number[1] for Camli blobs. As
JSON serialization libraries [sanely] won't let you specify the
order of keys in a dictionary, the camli serialization libraries
will wrap the JSON serialization libraries and replace the leading
'{' character with the above magic number.
Object Origin: for objects intended to be mutated over time
(e.g. annotating and renaming a photo, commenting on an item,
starring something), you need a base immutable object to reference
when hanging other immutable objects off. So whenever you want
to create a high-level mutable object, clients first create a
new, random dummy "Object Origin". e.g.:
{"camliVersion":1,
"type":"origin",
"originDate":"2010-06-20T18:07:14Z",
"noIndexRandom":"6157669e-df09-4107-a569-c876767d2117"}
Note that the only required fields are the "camliVersion" and "type"
fields, identifying this object as an origin hub. The "originDate"
is optional but recommended. It says when the origin object was
created (but not when any of the objects it binds were created).
The "noIndexRandom" field simply adds randomness to the object, such
that the BlobRef of the object is unique, even for two objects
created at the exact same time by different machines. You should
incorporate something specific from your machine (MAC address, IMEI,
etc) in your random calculation if you're not using a good
randomness source (e.g. using the "uuidgen" tool) The key name
"noIndexRandom" is not part of the spec, and it should definitely
not be indexed at higher levels. It's never referenced and not
semantically meaningful.
Object Origin Blob: the bytes of the JSON serialization of the Object
Origin. As it's just a blob, it's referred to using a BlobRef.
----------
References
----------
[1] http://en.wikipedia.org/wiki/Magic_number_(programming)
http://en.wikipedia.org/wiki/File_format#Magic_number