Why does Document need a lifetime?

[EverlastingBugstopper]

I'm trying to parse a schema into AST, and I'm wondering why the Document type it produces requires a lifetime. I'd like to be able to use the document as an owned struct without having to worry about lifetimes at all.

[tailhook]

We had that in previous implementations, but some users want parser with less memory allocations. So what you can do is parse_query<String>(...)?.into_static() to get a Document with static lifetime (which generally an equivalent of the thing without a lifetime in Rust).

[EverlastingBugstopper]

appreciate it!!

[mathstuf]

So this change is making the update for graphql_client_codegen absolutely ridiculous. I'm hours into chasing down lifetimes and mismatches and I have no idea how deep this rabbit hole will continue to go. Reborrowing in scopes doesn't convince the compiler, there are all kinds of method calls that borrow which require turning into owned data in order to use the types in the future, etc. I'll note that using 'static ends up unifying lifetimes in the wrong places (this is what I'm currently trying to untangle) meaning that other parameters end up being requested to be 'static. Was it really worth removing the non-lifetime variants?

[mathstuf]

Ok, so by just using 'static, String "everywhere" I was able to get to at least update to 0.3. I'll make another attempt to not do so everywhere next.

[tailhook]

I think it's fine to use 'static, String for codegen, as it's not performance critical part of the code.

(I'm sorry that it took you too much time, we probably may improve our docs suggesting that static lifetime is okay?)