Use the defaultName pattern for suggested model type names.

[stephenh]

There is technically an assumption in here that we should use the 1st file's defaultName, when the user might have N files with different defaultNames, but I think this will be what the user wants most of the time.

Fixes #226.

[Weakky]

Hey!

Thanks for your work again! I think we need to change the design a bit regarding this feature.

Always using the first one seems really arbitrary and will lead to other problems.
Once you have several files with different defaultName, choosing the first one doesn't make any sense anymore.

My suggestion is the following:

Use the defaultName to name the suggested models when all the files have the same defaultName, otherwise, fallback to regular name.

If there's only one file with a defaultName, it should work as well.

What do you think?

[stephenh]

@Weakky sure, that sounds like a reasonable heuristic. I'll update to do that.

[stephenh]

I noticed the conflicts; will rebase. Also, per the design-decision flag, fwiw my use case/workflow for wanting this was running graphqlgen on an existing schema and looking at "...well, now I have a whole bunch of names to fix". :-)

(Also while talking about step 1 in general, I think it'd be cool if graphqlgen just wrote the types to where they are supposed to go. As if there is >more than a single console of output show, the copy/paste is somewhat tedious. Granted, historically I was verify hesitant about code generators writing to files that are also programmer-maintained, in case of accidental overwrites, but it seems like jest's inline snapshot feature is perfectly happy to do that and people, even myself somewhat surprisingly, have been generally accepting of that workflow. Not a big deal, but FWIW. I can/should file as a separate issue too.)

[schickling]

@stephenh would you mind rebasing and resolve the conflicts? Let's merge then ✅

Re automatic model/type generation, let's discuss this over here: #181

[stephenh]

@schickling okay, should be good.