aboutsummaryrefslogtreecommitdiff
path: root/doc/api.md
Commit message (Collapse)AuthorAgeFilesLines
* use POST for requests with input dataLinus Nordberg2021-05-251-20/+32
| | | | | | | The major argument for moving input data from HTTP headers in GET requests to body of POST's is that we define the protocol ourselves without any dependencies on HTTP and can make it even simpler to parse.
* signatures are 64 octetsLinus Nordberg2021-05-111-1/+1
| | | | Spotted by Rasmus.
* use backticks for quoting single wordsLinus Nordberg2021-05-041-39/+39
| | | | I think this is more markdownish.
* another pass over the input and output descriptionsLinus Nordberg2021-05-041-64/+68
| | | | Mostly replacing "human-readable" with something more well defined.
* no need for encoding SPACELinus Nordberg2021-05-041-4/+4
|
* move long description of sharding to the design docLinus Nordberg2021-05-041-35/+14
|
* seconds, not millisecondsLinus Nordberg2021-05-041-1/+1
|
* whitespace changesLinus Nordberg2021-05-041-142/+181
|
* be consistent with "request" vs "entity" headersLinus Nordberg2021-05-041-1/+1
|
* explain how input and output data are sentLinus Nordberg2021-05-041-0/+10
| | | | | | | This is the "header in, body out" idea written up. We might change to a "POST body in, receive body out" scheme with "Content-Type: application/stfe" if we can decide that POST is not a terrible idea after all.
* specify serialization of keyLinus Nordberg2021-05-041-2/+1
|
* get rid of the underspecified term "ordinary users"Linus Nordberg2021-05-041-1/+2
|
* be explicit with key type; define struct message, for tree_leafLinus Nordberg2021-05-041-25/+34
| | | | | | | Specify who's verification key -- log, witness or submitter. Move shard_hint and checksum in tree_leaf into its own struct, for a more explicit definition of what to be signed.
* Incorporate changes from recent discussions.Linus Nordberg2021-05-031-85/+105
| | | | | | | | | | | | | | | | | | Remove all RSA support. Motivation: Simpler format for tree_leaf. Replace percent-encoding with headers for indata and key/value in body for outdata. Motivation: '<key>:<value>' is exactly what we want and it works for output data (responses) and not only for input data (requests). Don't POST. Motivation: We don't need the complexity of POST since we don't ever send a lot of data to the log. Split up the get-signed-tree-head endpoint into three separate without input data. Motivation: More explicit API plus easier debugging. Change timestamps and shard hints to use seconds rather than milliseconds. Motivation: time(1) and time(2).
* drafty experiment where we would only use percent encodingRasmus Dahlberg2021-04-261-90/+116
|
* moved shard_hint into tree_leafRasmus Dahlberg2021-04-261-51/+63
|
* added shard_hintRasmus Dahlberg2021-04-201-0/+34
|
* started on our in-progress (re)design documentsRasmus Dahlberg2021-04-201-0/+247
|
* removed old filesRasmus Dahlberg2020-10-201-205/+0
|
* renamed Item typeRasmus Dahlberg2020-10-201-20/+20
|
* removed timestamp complexityRasmus Dahlberg2020-10-201-15/+5
| | | | | The metrics that one could derive can also be exported by the operator. It is probably good enough and does not justify complexity right now.
* fixed typoRasmus Dahlberg2020-10-191-1/+1
|
* added public endpointsRasmus Dahlberg2020-10-151-9/+94
|
* updated data structure specificationRasmus Dahlberg2020-10-151-51/+80
| | | | | | - Repurposed the CT/bis TransItem structure - Added a "version 1 checksum" leaf - Moved signature into leaf appendix
* started filling in the blanksRasmus Dahlberg2020-10-141-7/+57
| | | | | | - Added a basic Merkle tree leaf data structure that is extensible - Added explicit references (where possible) to RFC 6962. More to come. - Added TODOs that we need to think more about.
* added start on st log api documentationRasmus Dahlberg2020-10-131-0/+51
generated by cgit v1.2.3 (git 2.39.1) at 2024-11-11 05:18:31 +0000