gqlparser v2.3.1

[StevenACoffman]

gqlparser v2.3.1 appears to work ok, but v2.3.0 gqlparser had a PR that needed to be reverted.

FYI I cut a new v2.3.0 gqlparser release https://github.com/vektah/gqlparser/ thinking it had been a really long time, but it looks like the vektah/gqlparser#112 might have been half baked as it causes problems for gqlgen and genqlient:

go get -d github.com/vektah/gqlparser/v2@v2.3.0
go get github.com/99designs/gqlgen/cmd@v0.14.0
go mod tidy
go generate ./...
go test ./...
$ make example

Signed-off-by: Steve Coffman steve@khanacademy.org

[benjaminjkraft]

Thanks for testing and fixing the issue! Is there a reason to upgrade in genqlient? If there's a new feature we need or whatnot I'm happy to but I don't know if there's a reason to make our users upgrade if not (they can always just choose to upgrade in their own go.mod).

[StevenACoffman]

Mostly vektah/gqlparser@v2.3.1 has bugfixes that end users may notice when using current genqlient out-of-the-box. I was looking at how hard it would be to implement #164 by fixing vektah/gqlparser#166 ( some initial work here: vektah/gqlparser#169 ) and I wanted to insure that future releases of vektah/gqlparser would not break genqlient.

What's Changed in gqlparser v2.3.1 since v2.2.0

• make importer reproducible by @vvakame in make reproducable importer vektah/gqlparser#112
• Update prelude to match graphql_spec for @deprecated directive by @danielnovograd in Update prelude to match graphql_spec for @deprecated directive vektah/gqlparser#155
• Stop validator mutating the provided *Schema by @xzyfer in Stop validator mutating the provided *Schema vektah/gqlparser#158
• feat: do not mark __typename field as unknown in the schema validation by @damianopetrungaro in feat: do not mark __typename field as unknown in the schema validation vektah/gqlparser#172
• Fix: argument comparison between two sets of args by @jufemaiz in Fix: argument comparison between two sets of args vektah/gqlparser#170
• Fix for update by reference of default variable values by @aurelijusbanelis in Fix for update by reference of default variable values vektah/gqlparser#161

[csilvers]

I feel ben should have the final word, but I'm fine with this fwiw.

[benjaminjkraft]

Thanks for the explanation. I thought about it more and you've convinced me -- since gqlparser is internal to genqlient it makes good sense for us to upgrade it even though users may also use it via other libraries (i.e. gqlgen). If it doesn't work right they see that as our bug even if the cause is in gqlparser.