remove extra call to packages.Load fix #2505

[vikstrous]

This removes a slow and seemingly unnecessary call to packages.Load

I don't know how to test this because it's not clear what it's solving and how to reproduce the issue. In our codebase nothing bad happens, it just runs faster.

I have:

• Added tests covering the bug / feature (see testing)
• [N/A] Updated any relevant documentation (see docs)

[coveralls]

Coverage: 75.529% (-0.04%) from 75.565% when pulling 098f337 on vikstrous:remove-extra-load into 9d22d98 on 99designs:master.

[StevenACoffman]

@vikstrous Can you verify that this does not cause an issue when following the example documentation and doing the go run github.com/99designs/gqlgen init ?

This code was added to fix #1765
in #1773

it is possible that it is not still necessary but I would like you to verify that. If it is necessary, perhaps there is a better way to fix the original problem that doesn't cause the redundant load for every other situation.

[vikstrous2]

The quick start steps work with and without this PR.

If I comment out the autobind line (in the gqlgen.yml in the generated example) and/or add one that points to a package that doesn't exist, generate fails. With and without this PR.

This is the error:

reloading module info
generating core failed: unable to load example/graph/model - make sure you're using an import path to a package that exists
exit status 1

There was a test added in the linked PR and all tests are passing on this PR. I don't know which test suite it's in and there are a lot of them, but I'm guessing that whatever tests were added there are still running.

I'm really curious about "or just call packages.Load() twice" from #1767. I don't understand how calling load twice changes anything. It seems like @ipfans knows something I don't?

---

Re: the other reloads... It seems like all the extra calls to load are related to changes to packages caused by mixing up input and output packages. I want to dig into this more and see if I can remove more calls to load. Source: #1762 Maybe just making all calls to "ReloadAllPackages" more specific will do what I want. I'll look at that next.

[StevenACoffman]

Great! I appreciate your digging into this!