IRC log for #rest, 2014-12-19

https://trygvis.io/rest-wiki/

All times shown according to UTC.

Time	Nick	Message
01:36		talios joined #rest
02:20		lufi joined #rest
03:03		lufi joined #rest
03:15		lemur joined #rest
03:53		tr3onlin_ joined #rest
04:03		lufi joined #rest
04:21		lufi joined #rest
04:43		tr3online joined #rest
05:03		lufi joined #rest
05:43		lufi joined #rest
06:17		lemur joined #rest
06:17		lemur joined #rest
06:28	Jarda	in my API I have this resource/route called /expenses
06:28	Jarda	and expenses can be added by POSTing to /expenses with a json object
06:28	Jarda	now I have to add bulk upload from various formats (csv/xml etc)
06:29	Jarda	I could of course use the same POST to /expenses and difference from content-type request header
06:29	Jarda	but then the response will be a problem
06:30	Jarda	now my responses are JSON objects with a 201 status code and Location header
06:30	Jarda	and if I bulk upload to the same resource then I can't return just one object nor a location header
06:31	Jarda	should I just create a new 'virtual' resource where to upload
06:31	Jarda	like POST /expenses/_batch ?
06:32	Jarda	and that could then just return something like {"message": "OK", "created_count": 142} ?
06:42		lufi joined #rest
07:12		lufi joined #rest
07:39		lufi joined #rest
08:05		lufi joined #rest
08:11		lemur joined #rest
08:17		proteusguy joined #rest
08:23		lufi joined #rest
08:38	trygvis	Jarda: that would work
09:02		tr3online joined #rest
09:04	trygvis	I think I would definitely lmake ma
09:04	trygvis	ops
09:04	trygvis	make a batch endpoint for something like that
09:04		marcoslamuria joined #rest
09:12		marcoslamuria joined #rest
09:21		graste joined #rest
09:23		lufi joined #rest
09:37		interop_madness joined #rest
10:19		lufi joined #rest
10:23		marcoslamuria joined #rest
10:36		marcoslamuria joined #rest
10:39		Left_Turn joined #rest
10:53	interop_madness	i don't seem to find a good way to map searching functionality to a REST resource. i could only come up with POST myserver/myresource/search (with search constraints inside the body), or GET /myserver/myresource/search?constraint1=value1&constraint2=value2... any better ideas?
11:03		tr3online joined #rest
11:10	trygvis	that's your main options
11:10	trygvis	but this all comes down to what your hypermedia can express
11:37	Jarda	I have done 'cached' searches
11:38	Jarda	POST /search with search terms in body
11:38	Jarda	it responds with a redirect to /search/134134
11:38	Jarda	and GETtin /search/134134 always returns the same results
11:38	Jarda	I hate it when going through paginated results and the results change between pages
11:39	Jarda	so I either miss some of the results or get sometimes doubles
11:40	trygvis	yeah. but I would default to the technique I talked about yesterday. don't use page=nn, use after_id=xyz
11:40	trygvis	it's the same thing as using the indexes to get fast paging in postgresql :)
11:41	Jarda	yeah I have never thought of that
11:41	Jarda	but then that won't work with ordered results
11:41	trygvis	sure it wil
11:41	trygvis	l
11:42	trygvis	I usually either sort by either id or lastUpdatedDate
11:42	Jarda	how about orderBy=name
11:43	Jarda	even with lastUpdatedDate you can't workaround the problem that you might miss some search results if the dataset changes
11:43	trygvis	how is that a problem?
11:43	trygvis	I also have createdDate so you can constrain it with it having a createdDate before <timestamp> too
11:44	Jarda	well maybe that's just my personal taste
11:44	Jarda	but I don't like that when (UI-wise) I go through paginated results I might miss some results
11:44	trygvis	this is a networked service so you will have to describe if stuff can change in your feeds
11:44	Jarda	or I get same results multiple times
11:44	trygvis	yeah, that is a problem
11:48	trygvis	if you want something perfect you will have to save the result set somewhere
11:48	Jarda	yes
11:48	trygvis	(note how all of these solutions shows how important it is that the server creates the URL!)
11:49	Jarda	I have done stuff where I save the search result set to redis
11:49	Jarda	and then POSTing to the result url will refresh the search
11:50	trygvis	what does the client post to get the update?
11:50	trygvis	sounds almost like a subscription more than a normal search
11:52	trygvis	couldn't the client just use one of those headers who's name I can think of now to request a fresh version?
11:54	Jarda	If-None-Match
11:54	Jarda	that's actually a very good question
11:54	Jarda	Why didn't I use ETag + If-None-Match :D
11:54	trygvis	no, that doesn't force a refresh. it just tells the client to reuse what it has
11:54	trygvis	but it's a good combination
11:54	Jarda	oh
11:54	Jarda	I got these mixed now
11:54	trygvis	http://tools.ietf.org/html/rfc2616#section-14.9
11:54	Jarda	mabe it's cache headers
11:54	Jarda	I would have to use
11:55	trygvis	min-fresh seems like what you want
11:55	trygvis	the upside is that you can use a generic http cache in front too
11:55	Jarda	mmm, yeah
12:38		Left_Turn joined #rest
12:47		lufi joined #rest
13:05		tr3online joined #rest
13:44		nkoza joined #rest
13:56		Mxyzpltk joined #rest
14:06		tr3online joined #rest
14:47		marcoslamuria joined #rest
15:07		tr3online joined #rest
16:16		quimrstorres joined #rest
17:08		marcoslamuria joined #rest
17:08		tr3online joined #rest
17:12		mezod joined #rest
17:20		proteusguy joined #rest
17:33		fumanchu_ joined #rest
17:54		proteusguy joined #rest
17:54		Mxyzpltk joined #rest
18:08		scflode joined #rest
18:30		rosstuck joined #rest
18:33		marcoslamuria joined #rest
18:36		rosstuck joined #rest
18:47		tr3online joined #rest
19:05		tmoore_ joined #rest
19:09		ekroon__ joined #rest
19:48		scflode joined #rest
19:59		graste joined #rest
20:01		Mxyzpltk joined #rest
20:05		rosstuck joined #rest
20:20		whartung joined #rest
20:23		graste joined #rest
21:09		graste joined #rest
21:20		graste joined #rest
21:30		graste joined #rest
22:41		proteusguy joined #rest

https://trygvis.io/rest-wiki/