<https://github.com/serverless-stack/serverless-st...
# sstö
Ömer Toraman
10/20/2021, 4:33 PMt
thdxr
10/20/2021, 4:34 PMI was actually thinking of inheriting from @tsconfig/node14 - it's what I do for all my projects
ö
Ömer Toraman
10/20/2021, 4:35 PMMe too, though I add some additional stuff.
However, instead of inherting embeding it would be better for more control
Ömer Toraman
10/20/2021, 4:36 PMPlus the default runtime for node defaults to node 12 when u create with create serverless stack.
Thus I guess it is better to inherit from tsconfig/node12
t
thdxr
10/20/2021, 4:38 PMYeah I thinkw e should update that too
thdxr
10/20/2021, 4:38 PMthere's a series of things I always do after I make a new project that should just be in the core template
ö
Ömer Toraman
10/20/2021, 4:46 PMI agree, I change the node version every time 😄
Ömer Toraman
10/20/2021, 4:48 PMYeah me too 😄
Ömer Toraman
10/20/2021, 4:52 PMCan we pass the StackName?
Ömer Toraman
10/20/2021, 4:52 PM%stack-name.PascalCased%.template.tss
Simon Reilly
10/20/2021, 6:50 PMI always change from default export to named export, but maybe that's bike shedding
Simon Reilly
10/20/2021, 6:53 PM@Ömer Toraman I worked on a command for this, since I am always doing multi stacks. https://github.com/simonireilly/serverless-stack/pull/4
Copy code
sst add-stack MyStackNameö
Ömer Toraman
10/20/2021, 6:55 PMYes! I agree that using named export matters for both TypeScript and VS Code. I try not to use default exports as much as possible.
Ömer Toraman
10/20/2021, 6:55 PMI like the idea! I previously used nestjs, and honestly like how it creates some resources with CLI
s
Simon Reilly
10/20/2021, 6:57 PMYeah, all the templates are there already tbh. Just case of plugging it together.
Also, there is documentation to show how to use multistacks, but better to have CLI. This follows show, don't tell, principals
ö
Ömer Toraman
10/20/2021, 7:08 PMThis is not the part of this discussion, but what I’m looking for is to really take the pain away inside Lambda functions.
deserializing, validation, serializing, logging, error handling
It quickly becomes one big function that is hard to read and a mess.
s
Simon Reilly
10/20/2021, 7:14 PMWonder about this a lot too. Especially with frameworks like fastify, and tsoa in server bound deployments.
Still, the contract between the application type, and the openAPI schema is something I think works for these deserializing, validation, serialising and error handling definition.
Logging itself is a separate concern IMO. I just bang in pino logger for JSON logs.
ö
Ömer Toraman
10/20/2021, 7:46 PMDo u use openAPI schema?
Ömer Toraman
10/20/2021, 7:47 PMBut yeah it is great
Ömer Toraman
10/20/2021, 7:47 PMBut how would you know that the request was rejected because of validation for example, since it will never hit the Lambda
Ömer Toraman
10/20/2021, 7:53 PMWhen I was looking for a logger, I saw this package made with Pino for lambda:
https://www.google.com/search?q=pino+lambda&oq=pino+lambda&aqs=chrome..69i57j0i10l2j0i10i512j0i10l6.1476j0j4&sourceid=chrome&ie=UTF-8
Ömer Toraman
10/20/2021, 8:04 PMPino says it is 5 times faster, but the benchmark they show doesnt seem so lol
Ömer Toraman
10/20/2021, 8:04 PMs
Simon Reilly
10/21/2021, 7:57 AMCouple of questions there, one, pinonis probably fast enough for me, especially since we aren't too worried about concurrency in lambda.
Second, yeah, I wouldn't integrate open API with aws apigateway rest API v1. To me that's too much buying into AWS, would rather keep the app portable. Something like ajv JSONSchemaType<T> with validator and serializer is what I'd be looking at
ö
Ömer Toraman
10/21/2021, 10:14 AMI see. I use class validator instead.
Ömer Toraman
10/21/2021, 10:15 AMAny microservice framework do u use, if it can be thought as a framework
Ömer Toraman
10/21/2021, 10:15 AMlike middy
Ömer Toraman
10/21/2021, 10:15 AMor this
Ömer Toraman
10/21/2021, 10:15 AM2 Views