Difference between revisions of "Talk:Ackis2.0/Protocol"

From Makers Local 256
Jump to: navigation, search
(discussion on protocol additions including auth and general information querying involving the core)
 
(add discussion on component transactions)
 
(One intermediate revision by one user not shown)
Line 1: Line 1:
 +
== auth ==
 
I don't know how we want to do authentication for remote (or local) users.  I'd imagine it would use a hashing function like MD5 or SHA1 on a concatenation of the responseid, username, and password in a defined order.  Probably an "authenticate" tag. Instead of implementing a specific "component" tag, we should probably create a more generic "information" tag in which you can query the core for data about its internals. --[[User:Opticron|Opticron]] 22:49, 10 July 2008 (CDT)
 
I don't know how we want to do authentication for remote (or local) users.  I'd imagine it would use a hashing function like MD5 or SHA1 on a concatenation of the responseid, username, and password in a defined order.  Probably an "authenticate" tag. Instead of implementing a specific "component" tag, we should probably create a more generic "information" tag in which you can query the core for data about its internals. --[[User:Opticron|Opticron]] 22:49, 10 July 2008 (CDT)
 +
 +
== resources ==
 +
 +
Are resources allowed to utilize other resources?  This would prevent resources from being included in the message family of transactions.  I think it should be its own tag anyway, because the transactions serve a fundamentally different purpose from typical messages.
 +
 +
== component transactions ==
 +
Currently, it is specified that the core must support action=list with type=actions and type=types.  A better convention would be to only require action=list with type=actions.  All other meta-info queries to get information about a specific action should be requested using action=list with type=<action>.  As such, to find out what types you can use with the list action, you would do action=list with type=list and get back "list,actions" at bare minimum.  A more complete core might respond to the same query with "actions,list,components,modules,clients,resources,connections". --[[User:Opticron|Opticron]] 00:46, 26 April 2009 (CDT)

Latest revision as of 00:46, 26 April 2009

auth

I don't know how we want to do authentication for remote (or local) users. I'd imagine it would use a hashing function like MD5 or SHA1 on a concatenation of the responseid, username, and password in a defined order. Probably an "authenticate" tag. Instead of implementing a specific "component" tag, we should probably create a more generic "information" tag in which you can query the core for data about its internals. --Opticron 22:49, 10 July 2008 (CDT)

resources

Are resources allowed to utilize other resources? This would prevent resources from being included in the message family of transactions. I think it should be its own tag anyway, because the transactions serve a fundamentally different purpose from typical messages.

component transactions

Currently, it is specified that the core must support action=list with type=actions and type=types. A better convention would be to only require action=list with type=actions. All other meta-info queries to get information about a specific action should be requested using action=list with type=<action>. As such, to find out what types you can use with the list action, you would do action=list with type=list and get back "list,actions" at bare minimum. A more complete core might respond to the same query with "actions,list,components,modules,clients,resources,connections". --Opticron 00:46, 26 April 2009 (CDT)