In the examples there is a `resolvers.ts` file in ...
# orm-helpk
kitze
09/28/2018, 5:59 AMIn the examples there is a
resolvers.tsgeneratedprisma deployprisma generatekitze
09/28/2018, 7:35 AMGot it
graphql-resolver-codegenkitze
09/28/2018, 7:36 AMBut I'm still not sure when should this run. After every deploy?
b
brikou
09/28/2018, 7:42 AMyou should run it after each changes on schema.gql (or datamodel.gql as inherited)
brikou
09/28/2018, 7:42 AManyway don't use last version because bug in npmignore, so some files are missing under ./generetors
k
kitze
09/28/2018, 7:45 AMYeah I switched to a fork
kitze
09/28/2018, 7:45 AMBut once the
/resolverskitze
09/28/2018, 7:45 AMso you basically generate the resolvers scaffold once, and then you only generate the interfaces?
b
brikou
09/28/2018, 7:54 AMyou can use --force flag and then
git checkout -p -- resolversDirbrikou
09/28/2018, 7:55 AMnot optimal at this time, more close to a POC but nice anyway
k
kitze
09/28/2018, 7:56 AM😐
kitze
09/28/2018, 7:56 AMIt would be cool to add new "empty" methods for the unimplemented resolvers
kitze
09/28/2018, 7:56 AMInstead of rewritting the files that already have some defined resolvers
b
brikou
09/28/2018, 7:59 AMmore than empty, generated code should be patch using TS ast 🙂
brikou
09/28/2018, 8:00 AMin fact, I think it already work the way you describe for new types, not for fields
k
kitze
09/28/2018, 8:04 AMThere is so much potential for auto-codegen with Prisma + TS
👍 2
d
divyendu
09/28/2018, 8:34 AMI will clean this up (the npmignore, build stuff) today and ping here 🙂
A few notes on the design decisions
- For scaffolding, it is a 1 time process and without the -f flag we still rewrite types i.e. making TS compiler complaint about the new fields and you have to expand them manually
- We thought about using TS AST and adding missing resolvers but generally whoever we spoke to suggested against it.. almost all scaffolders don't edit existing files.. we can change the folder structure to make resolvers their own files and add new files but that would be too many files.
All of this is our current thought process (documented at the end of readme I think) but these are all awesome and valid points. Please feed free to raise issues for them to further discuss as slack discussions are likely to get lost 🙏
k
kitze
09/28/2018, 8:37 AMWhat you're saying makes sense
kitze
09/28/2018, 8:37 AMBut adding an additional field, for example "surname" to a model, and then editing 30 different files is just exhausting
d
divyendu
09/28/2018, 8:43 AMI totally understand that.. we are sort of tracking this via the 1st point here: https://github.com/prisma/graphql-resolver-codegen/issues/16
but feel free to raise a separate issue for this.. we would love to think about solutions on these lines 🙂
divyendu
09/28/2018, 9:35 AMI cleaned it up and released a fix for the build (includes etc).
0.2.4@betak
kitze
09/28/2018, 10:13 AMGreat, tnx!
kitze
09/29/2018, 3:57 AMI gave up on using the graphql-resolver codegen ...
d
divyendu
09/29/2018, 8:07 AMIf you can document all you feedback in form of an issue, that would help. Thanks!
3 Views