sysutils/locki: Fix checksum

[grembo]

Summary

This reverts commit 1dff87feb9cdb23f040422deba3697f5cebf79dc.

The upstream file was restored to its previous version.

This fixes checksum locally for me. Please confirm it's not limited to my system. Could be some regional CDN problem.

Convert the port to pure go:modules (GH_TUPLE is not required), which is a) better and b) solves the checksum issue as a side-effect.

Test Plan

make checksum in the port

Build and install the port, poudriere testport.

[adamw]

Hate to say it, but make distclean && make checksum succeeded for me as-is.

[grembo]

@bdrewery @adamw Well, so for me (from Germany in various locations/ISPs) buildings works with the patch I committed and would fail without it (just tried on a new server, since the package wasn't available).

On the other hand, builds on the cluster are failing now (which I assume is US of A).

So might really be a CDN issue. Any way this could be solved (we don't allow multiple alternative checksums per file, do we)?

fetch -qo - https://codeload.github.com/hashicorp/consul/tar.gz/v1.5.1?dummy=/hashicorp-consul-v1.5.1_GH0.tar.gz | sha256
704511c916beb2e3de5cb4fc5d183d6e37bb31ad78252acd20da8a9e1153ebe0

[adamw]

Reordering the quoted reply for dramatic effect.

In D55695#1280564, @grembo wrote:

fetch -qo - https://codeload.github.com/hashicorp/consul/tar.gz/v1.5.1?dummy=/hashicorp-consul-v1.5.1_GH0.tar.gz | sha256
704511c916beb2e3de5cb4fc5d183d6e37bb31ad78252acd20da8a9e1153ebe0

I get that same output. And to confirm, that's what's currently in the tree, correct?

So might really be a CDN issue. Any way this could be solved (we don't allow multiple alternative checksums per file, do we)?

So, I'm unclear why this port is using both USES=go:modules and GH_TUPLE. The whole point of go:modules is that you don't need to list out stuff in GH_TUPLE.

My advice is: remove GH_TUPLE and post-extract:, re-run make makesum, and you should be good to go.

[grembo]

@adamw

I get that same output. And to confirm, that's what's currently in the tree, correct?

That is correct, and since the only difference is a longer hash inside the archive, it should be the "correct" one long term (I don't think they will shorten hashes).

My advice is: remove GH_TUPLE and post-extract:, re-run make makesum, and you should be good to go.

I'll try that (the maintainer seems to have only this one port and isn't in phabricator).

[grembo]

Upload patch, so review matches what we discussed.

[grembo]

@bdrewery Thanks for reporting
@adamw Thanks for your help