Post 9tnHu1WNzKN2d7p9SS by Ark@linuxrocks.online
(DIR) More posts by Ark@linuxrocks.online
(DIR) Post #9tmExiKOnRPVEzKEnw by sir@cmpwn.com
2020-04-06T19:51:10Z
0 likes, 0 repeats
Took a crack at designing a GraphQL schema for git.sr.ht, comments welcomehttps://paste.sr.ht/~sircmpwn/94578e95691aac2047bb0703520e7fb8c53d6de9
(DIR) Post #9tmF6zbdvTrJQyod0q by lain@lain.com
2020-04-06T19:53:31.523740Z
0 likes, 0 repeats
@sir what do you think about graphql? I would not have expected you to use it.
(DIR) Post #9tmFKiEhIw9bcFcotc by sir@cmpwn.com
2020-04-06T19:55:13Z
1 likes, 0 repeats
@lain I think it's an substantial improvement on the typical REST kludge, but it:1. Has footguns2. Has mostly poor quality server-side implementations3. Leaves a lot of questions to the implementor which probably should have been answered upstream
(DIR) Post #9tmHXLPDhKtITwhKro by riking@social.wxcafe.net
2020-04-06T20:19:36Z
0 likes, 0 repeats
@sir my #1 feedback is "add more comments", they render as documentation. Document every type and every field even if you feel daft, except for Query and Mutation. https://github.com/2fd/graphdoc
(DIR) Post #9tmHdXuM7zxkziDci8 by sir@cmpwn.com
2020-04-06T20:20:01Z
0 likes, 0 repeats
@riking ack, but that's not important for the prototype
(DIR) Post #9tmHwmQUNDvpWYQ7Hs by riking@social.wxcafe.net
2020-04-06T20:24:28Z
0 likes, 0 repeats
@sir like i have no clue what a RepoInput is right now, so I can't really offer feedback on it.PGPSignature should probably reference the key performing the signing and the server's evaluation of the signature's trust status. Or you can make the decision that `valid:false` is just a footgun and users need to extract invalid signatures from the raw git object.
(DIR) Post #9tmMSmIamYciDEEw9Q by tomleb@toot.cafe
2020-04-06T21:14:41Z
0 likes, 0 repeats
@sir Is the goal here to replace the current api? Or to maintain both? Or just for fun?
(DIR) Post #9tmMZaxzJWTt5oqxGK by sir@cmpwn.com
2020-04-06T21:15:24Z
0 likes, 0 repeats
@tomleb likely to replace the current API. It's been subject to change since the start, being an alpha, and now that more of the services are developed the limitations of the current design are apparent and offer constraints on a possible final design
(DIR) Post #9tmO6GzMiBQydwdwMy by tomleb@toot.cafe
2020-04-06T21:33:21Z
0 likes, 0 repeats
@sir I haven't been involved much in the project, I'd be interested to know what the limitations of the current design is. (Might be a good idea for a technical blogpost, although I'm sure you already have a big list of things to blog about)I did use the API at some point and there were differences between the information from the web pages vs. received from the API. Would that be fixed?
(DIR) Post #9tmOCAenAMRqk7pg7E by sir@cmpwn.com
2020-04-06T21:33:56Z
0 likes, 0 repeats
@tomleb no, that wouldn't be fixed. The main issue is inconsistencies between APIs, and limitations in what kind of data you can access and how it's organized
(DIR) Post #9tmVq158stQ4JE6XsO by val@oc.todon.fr
2020-04-06T22:58:03Z
0 likes, 0 repeats
@sir Not sure if you care, but I think it's missing committer time, and a time for tags
(DIR) Post #9tmVwbkNTepXlGXEnI by sir@cmpwn.com
2020-04-06T22:58:37Z
0 likes, 0 repeats
@val ty
(DIR) Post #9tnHu1WNzKN2d7p9SS by Ark@linuxrocks.online
2020-04-07T07:58:06Z
0 likes, 0 repeats
@sir There's a typo on line 216: visibiilty instead of visibility
(DIR) Post #9tnbBqDmddm7vQKgJU by sir@cmpwn.com
2020-04-07T11:34:54Z
0 likes, 0 repeats
@Ark cheers
(DIR) Post #9tqBdpiI6c5f7lAXCK by tristan957@mastodon.social
2020-04-08T17:32:26Z
0 likes, 0 repeats
@sir @lain Quality server implementations are a real pain to find in a lot of languages. The best I have found are in the TypeScript ecosystem.JavaScript sucks.