graphql/yoga-server/docs/features/parsing-and-validation-caching

[giscus[bot]]

graphql/yoga-server/docs/features/parsing-and-validation-caching

By default, Yoga will maintain a parsing and validation cache. If requests contain documents that have been executed before, they will not be parsed and validated again.

https://the-guild.dev/graphql/yoga-server/docs/features/parsing-and-validation-caching

[AdiRishi]

How can we use an async cache? E.g I want to use Cloudflare workers KV within my own cache object, but the get and put methods are async.

[ardatan]

Unfortunately it is not possible currently because parse and validate from graphql-js are sync.

[AdiRishi]

Damn, ok fair enough. Hopefully one day these can be made async.

Although I'm not sure if it's faster to just have the server run the validation / parsing checks. I don't have a concept of how much time they take.
From what I know the KV.get/put methods are around 30 - 100ms.

[EmrysMyrddin]

Then it is probably not worth it. The parsing and validation are not that expensive. It's always good to cache it in memory, because it's an easy optimisation, but I don't think it is worth using an external tool for this.

You can have a sense of what takes time by enabling debug log levels :-)

[randomiser]

+1 for an async version.

I'm using AppEngine standard and in memory LRU caching isn't really supported. Would need to use Redis to get the parsing/validation performance gains.

[EmrysMyrddin]

As said, this is probably not going to happen in a short term, since it's a limitation of graphql-js. The only way to overcome this is to actually replace graphql-js with another engine, which we don't actually have nor support today.