Comments on: A method for massively cutting UPDATE/HTTP-PUT methods in complex Rest APIs http://themineproject.org/2009/01/a-method-for-massively-cutting-updatehttp-put-methods-in-complex-rest-apis/ open source project for online data and relationships logistics Fri, 18 Jun 2010 19:44:55 -0400 http://wordpress.org/?v=2.8.4 hourly 1 By: alecm http://themineproject.org/2009/01/a-method-for-massively-cutting-updatehttp-put-methods-in-complex-rest-apis/comment-page-1/#comment-134 alecm Mon, 02 Mar 2009 21:40:18 +0000 http://themineproject.org/?p=218#comment-134 Hi PJ - thanks for the input >* the /key/ bit seems redundant. What’ s wrong with just /42/keyName.xxx ? Clashes with other URL/operations in the same space, and I don't want to get into parsing a token and then guessing what it meant to imply and/or having to forbid certain keynames / parameter names because they clash with API calls... Love the batch-update point, I was wondering about that and you've helped me decide. - alec Hi PJ – thanks for the input

>* the /key/ bit seems redundant. What’ s wrong with just /42/keyName.xxx ?

Clashes with other URL/operations in the same space, and I don’t want to get into parsing a token and then guessing what it meant to imply and/or having to forbid certain keynames / parameter names because they clash with API calls…

Love the batch-update point, I was wondering about that and you’ve helped me decide.

- alec

]]> By: PJ http://themineproject.org/2009/01/a-method-for-massively-cutting-updatehttp-put-methods-in-complex-rest-apis/comment-page-1/#comment-133 PJ Mon, 02 Mar 2009 18:56:46 +0000 http://themineproject.org/?p=218#comment-133 Some thoughts: * the /key/ bit seems redundant. What' s wrong with just /42/keyName.xxx ? * Consider implementing UPDATE for aggregations anyway, if it's not too difficult. It will likely be useful to be able to atomically UPDATE a whole group of data in one swell foop, instead of having to individually set all those smaller pieces. Some thoughts:

* the /key/ bit seems redundant. What’ s wrong with just /42/keyName.xxx ?

* Consider implementing UPDATE for aggregations anyway, if it’s not too difficult. It will likely be useful to be able to atomically UPDATE a whole group of data in one swell foop, instead of having to individually set all those smaller pieces.

]]> By: Ben http://themineproject.org/2009/01/a-method-for-massively-cutting-updatehttp-put-methods-in-complex-rest-apis/comment-page-1/#comment-84 Ben Fri, 16 Jan 2009 23:47:58 +0000 http://themineproject.org/?p=218#comment-84 p.s. if you're using Java on the server-side look at open-source products like XStream (xstream.codehaus.org) for tools that can help with simple de-serialization of XML without the overhead of SOAP. p.s. if you’re using Java on the server-side look at open-source products like XStream (xstream.codehaus.org) for tools that can help with simple de-serialization of XML without the overhead of SOAP.

]]> By: Ben http://themineproject.org/2009/01/a-method-for-massively-cutting-updatehttp-put-methods-in-complex-rest-apis/comment-page-1/#comment-83 Ben Fri, 16 Jan 2009 23:44:55 +0000 http://themineproject.org/?p=218#comment-83 Thanks to the unstructured nature of the content of an HTTP POST you can aggregate updates of an object's values into a single request aimed at the object instead of the individual keys by using an xml/json object in an HTTP POST (or even just a set of key-value pairs!) and have a ReSTful server figure out where to apply them. That's not an abuse of the ReST concept as you're still only using the CRUD operations, just on larger objects than just strings and numbers. ReST is a nice design direction - we've been doing it for a couple of years (at least our non-purist form of it at my current place of work) and it's proved to be flexible enough to stand up to some quite diverse requirements (and it suits ExtJS single-page applications!). My only regret is that we chose to specify the representation as an additional parameter in the request (/api/object/42?rep=xml), rather than as the dot extension (/api/object/42.xml) as you have it. Thanks to the unstructured nature of the content of an HTTP POST you can aggregate updates of an object’s values into a single request aimed at the object instead of the individual keys by using an xml/json object in an HTTP POST (or even just a set of key-value pairs!) and have a ReSTful server figure out where to apply them. That’s not an abuse of the ReST concept as you’re still only using the CRUD operations, just on larger objects than just strings and numbers.

ReST is a nice design direction – we’ve been doing it for a couple of years (at least our non-purist form of it at my current place of work) and it’s proved to be flexible enough to stand up to some quite diverse requirements (and it suits ExtJS single-page applications!). My only regret is that we chose to specify the representation as an additional parameter in the request (/api/object/42?rep=xml), rather than as the dot extension (/api/object/42.xml) as you have it.

]]> By: dropsafe : A method for massively cutting UPDATE/HTTP-PUT methods in complex Rest APIs http://themineproject.org/2009/01/a-method-for-massively-cutting-updatehttp-put-methods-in-complex-rest-apis/comment-page-1/#comment-82 dropsafe : A method for massively cutting UPDATE/HTTP-PUT methods in complex Rest APIs Fri, 16 Jan 2009 09:05:16 +0000 http://themineproject.org/?p=218#comment-82 [...] The Mine! project » A method for massively cutting UPDATE/HTTP-PUT methods in complex Rest APIs [...] [...] The Mine! project » A method for massively cutting UPDATE/HTTP-PUT methods in complex Rest APIs [...]

]]>