| Time |
S |
Nick |
Message |
| 00:02 |
|
|
fuzzyhorns joined #rest |
| 00:48 |
|
|
fuzzy_horns joined #rest |
| 01:47 |
|
|
ralphschindler joined #rest |
| 02:36 |
|
|
fuzzy_horns joined #rest |
| 03:31 |
|
|
fumanchu_ joined #rest |
| 03:49 |
|
|
blahdeblah_ joined #rest |
| 04:20 |
|
|
ralphschindler joined #rest |
| 04:25 |
|
|
fuzzyhorns joined #rest |
| 04:52 |
|
|
baweaver joined #rest |
| 05:06 |
|
|
_ollie joined #rest |
| 05:44 |
|
|
ralphschindler joined #rest |
| 06:09 |
|
|
Left_Turn joined #rest |
| 06:14 |
|
|
fuzzyhorns joined #rest |
| 06:22 |
|
|
alxbl joined #rest |
| 07:15 |
|
|
interop_madness joined #rest |
| 07:31 |
|
|
fuzzyhorns joined #rest |
| 07:40 |
|
|
dEPy joined #rest |
| 08:02 |
|
|
graste joined #rest |
| 08:26 |
|
|
rosstuck joined #rest |
| 08:31 |
|
|
fuzzyhorns joined #rest |
| 08:41 |
|
|
whartung_ joined #rest |
| 09:50 |
|
|
vanHoesel joined #rest |
| 10:20 |
|
|
fuzzyhorns joined #rest |
| 11:52 |
|
|
eschmidbauer joined #rest |
| 11:53 |
|
|
eschmidbauer left #rest |
| 12:09 |
|
|
fuzzyhorns joined #rest |
| 13:04 |
|
|
rosstuck joined #rest |
| 13:10 |
|
|
fuzzyhorns joined #rest |
| 13:23 |
|
|
ralphschindler joined #rest |
| 13:38 |
|
|
fuzzyhorns joined #rest |
| 14:15 |
|
|
saml joined #rest |
| 14:23 |
|
|
nkoza joined #rest |
| 14:31 |
|
|
ralphschindler joined #rest |
| 14:52 |
|
|
FreezingDroid joined #rest |
| 14:53 |
|
|
baweaver joined #rest |
| 15:13 |
|
|
fuzzyhorns joined #rest |
| 15:19 |
|
|
foist joined #rest |
| 16:09 |
|
|
ralphschindler joined #rest |
| 17:18 |
|
|
sfisque joined #rest |
| 17:45 |
|
|
Rotaerk joined #rest |
| 17:45 |
|
Rotaerk |
hmm, don't really understand why HAL embeds links to documentation about the service, within the representation itself |
| 17:46 |
|
Rotaerk |
is the service meant to be consumable by the *developer* or by the *client* |
| 17:46 |
|
Rotaerk |
the client has no use for documentation, but the developer doesn't need the documentation to be provided inside the representations themselves... |
| 17:47 |
|
fumanchu_ |
I've found docs in representations to be hugely beneficial as a developer |
| 17:48 |
|
fumanchu_ |
much easier than matching up my current state with some wiki |
| 17:49 |
|
Rotaerk |
hmm, okay, so having the documentation attached in some way to the resources themselves helps to find it more easily... though it still seems like bloat that's unnecessary for the client |
| 17:50 |
|
Rotaerk |
seems like it'd be better to make a separate ... documentation media type |
| 17:50 |
|
Rotaerk |
if you want the documentation for the resource, you request that media type; but if you're the client, you just request the actual data |
| 17:50 |
|
Rotaerk |
data's media type* |
| 17:51 |
|
Rotaerk |
the documentation media type could just be text/html |
| 17:54 |
|
fumanchu_ |
better how? |
| 17:54 |
|
Rotaerk |
less bloat; less data to send |
| 17:54 |
|
Rotaerk |
if the client doesn't need it, then why send it |
| 17:54 |
|
whartung |
actually, if you want to do this, simply add a link in the resource to the documentation. |
| 17:55 |
|
Rotaerk |
afterall, a program can't read documentation |
| 17:55 |
|
whartung |
it's never expected too |
| 17:55 |
|
|
_ollie joined #rest |
| 17:55 |
|
fumanchu_ |
bloat hasn't been a problem for us in practice (and we're ostensibly doing HPC) |
| 17:55 |
|
whartung |
HPC? |
| 17:55 |
|
fumanchu_ |
high-performance computing |
| 17:55 |
|
fumanchu_ |
Big Data |
| 17:55 |
|
whartung |
ah k |
| 17:57 |
|
Rotaerk |
so basically, build the service to serve both the developer *and* the client-software with the same media type because there's no practical negative impact of doing so |
| 17:58 |
|
whartung |
Not really sure how a JSON based hypermedia (for example) really suits a developer. |
| 17:58 |
|
Rotaerk |
whartung: that's what I initially said, if you scroll up |
| 17:58 |
|
Rotaerk |
but fumanchu_ indicated otherwise |
| 17:59 |
|
Rotaerk |
whartung: if the JSON hypermedia isn't for the developer... if it's *just* for the client-software... then there's no point in linking documentation |
| 17:59 |
|
Rotaerk |
because the client-software can't read it |
| 17:59 |
|
Rotaerk |
so if you're linking it, clearly you're building it for the developer *and* the client software |
| 18:00 |
|
fumanchu_ |
that's a reasonable summary |
| 18:01 |
|
whartung |
that true to a point, but it sure is friendly, as a developer, to do a get on a resource and see a link to the relevent docs. You'll note that the bulk of the hypermedia documents are more or less "human readable", which offers no advantage to a client, but is mostly agreed upon (as a general theme) to be "a good thing"/ |
| 18:02 |
|
fumanchu_ |
often, the natural-language docs tell you what you can do next, and when you can actually express that in machine-readable structures, that's great! |
| 18:03 |
|
fumanchu_ |
there's a continuum from all-machine-readable -- partly machine-readable but partly dev-readable -- only dev-readable -- HTML for when "the machine" is a human driving the application |
| 18:03 |
|
Rotaerk |
yea HTML basically just dumbly forwards the responsibility of interpretation to the user |
| 18:04 |
|
whartung |
all formats do that lol |
| 18:04 |
|
whartung |
they're just bits for crying out loud. |
| 18:05 |
|
Rotaerk |
no, other formats must provide semantics that allows for client-software logic to understand |
| 18:05 |
|
whartung |
HTML has semantics |
| 18:05 |
|
Rotaerk |
HTML doesn't; it just describes these semantics in the human-language of choice |
| 18:05 |
|
Rotaerk |
as in, the link relations can be anything; they don't need to be standardized |
| 18:06 |
|
Rotaerk |
they just need to be something that would make sense to the person reading the page |
| 18:06 |
|
Rotaerk |
whereas for a service to be consumed by client-software, they need to be standardized... so the client can be like "if relation == "blah", then do this..." |
| 18:07 |
|
whartung |
actually people never see link relations, they see chrome and buttons and text. The fact that you can use non-standard link relations in HTML doesn't make it a bad media type. You're perfectly capable of using standard link relations in HTML |
| 18:25 |
|
Rotaerk |
reading an article about hypermedia formats... but it's from March 2014 |
| 18:25 |
|
whartung |
has't changed a whole lot in a year |
| 18:25 |
|
Rotaerk |
it covers JSON-LD, HAL, Collection+JSON, and SIREN |
| 18:25 |
|
Rotaerk |
any other major players I should know about? |
| 18:26 |
|
whartung |
I've not heard of SIREN |
| 18:26 |
|
Rotaerk |
it seemed to be an honorable-mention in the article; wasn't one of the recommended ones |
| 18:26 |
|
whartung |
I know mamund is working on Uber |
| 18:29 |
|
Rotaerk |
hmm wonder if that's supposed to be a progression of his Collection+JSON, or for a different purpose altogether |
| 18:31 |
|
Rotaerk |
currently debating between HAL and Collection+JSON |
| 18:32 |
|
fuzzyhorns |
Rotaerk: what are you wanting to do? |
| 18:32 |
|
fuzzyhorns |
Rotaerk: my impression is HAL is stripped down, so you will be rolling more of your own semantics on top of it than Collection+JSON |
| 18:32 |
|
fuzzyhorns |
so I favor the latter |
| 18:32 |
|
fuzzyhorns |
but I can see times when itd be overkill |
| 18:33 |
|
Rotaerk |
k |
| 18:33 |
|
fuzzyhorns |
Rotaerk: UBER covers all hypermedia controls, if you dont need all hfactors, i think Collection+JSON is fine |
| 18:35 |
|
Rotaerk |
interesting; thanks |
| 18:35 |
|
Rotaerk |
think I'll go with Collection+JSON for now |
| 18:39 |
|
* fumanchu_ |
has to at least mention http://www.bytebucket.org/fumanchu/shoji/src/f11eb85f65cbced875674ccfb283f8f08ff47e0d/spec.txt?at=default even though you won't adopt it |
| 18:40 |
|
Rotaerk |
is that your own or something? |
| 18:40 |
|
Rotaerk |
nm, I see your name in it |
| 18:40 |
|
fumanchu_ |
yeah, started it at Etsy, used it at a couple others since then (including current) |
| 18:55 |
|
|
adelphospro joined #rest |
| 19:05 |
|
|
adelphospro left #rest |
| 19:55 |
|
|
Andre-B joined #rest |
| 19:57 |
|
|
_ollie joined #rest |
| 20:26 |
|
|
fsvehla joined #rest |
| 20:32 |
|
|
sfisque joined #rest |
| 21:06 |
|
|
Coldblackice joined #rest |
| 21:13 |
|
|
sfisque1 joined #rest |
| 21:15 |
|
|
sfisque2 joined #rest |
| 21:44 |
|
|
sfisque joined #rest |
| 22:03 |
|
|
fuzzyhor_ joined #rest |
| 22:31 |
|
|
fuzzyhorns joined #rest |
| 23:11 |
|
|
fuzzy_horns joined #rest |
