Skip to content

fix(scaleway): a comment about the gateway's 404 outlived the line that fixed it - #783

Merged
stephrobert merged 1 commit into
mainfrom
fix/a-stale-comment-about-the-gateway-404
Sep 15, 2026
Merged

stephrobert merged 1 commit into
mainfrom
fix/a-stale-comment-about-the-gateway-404

Conversation

@stephrobert

Copy link
Copy Markdown
Owner

TestTheGatewayMetadataDoesNotVary carried a note saying that a POST to
/metadata answers net/http's plain-text 404 page not found, that the real
gateway answers a Scaleway error document there, and that the divergence was
left to #776.

All three sentences were true when they were written, and the first was false by
the time they were committed: declaring /metadata in productPrefixes — in
the same pull request
— is exactly what stops that path falling through to
net/http. The comment described a state its own change had removed.

Measured both ways on 2026-09-15:

answer
without the prefix 404 page not found (text/plain)
with it 501 {"type":"not_emulated", …}

What changes

The assertion is strengthened rather than the comment reworded. It was
!= 200, which passes for a bare status the SDK drops; it now names the refusal
this pack owes a client inside its declared space — which is #74's finding and
the reason the prefix list exists at all.

The existing mutation on that prefix reddens TestEveryRouteFallsUnderADeclaredPrefix,
a statement about the route table. This adds the one that reddens the
behaviour a client sees
. Seven mutations now, all biting.

mise run prepush green.

🤖 Generated with Claude Code

…at fixed it

TestTheGatewayMetadataDoesNotVary carried a note saying that a POST to
`/metadata` answers net/http's plain-text "404 page not found", that the real
gateway answers a Scaleway error document there, and that the divergence was
left to #776.

All three sentences were true when they were written and the first was false by
the time they were committed: declaring `/metadata` in productPrefixes — in the
same change — is exactly what stops that path falling through to net/http. The
comment described a state its own pull request had removed.

Measured both ways on 2026-09-15:

    without the prefix   404 page not found        (text/plain)
    with it              501 {"type":"not_emulated", …}

So the assertion is strengthened rather than the comment reworded. It was
"not 200", which passes for a bare status the SDK drops; it now names the
refusal this pack owes a client inside its declared space, which is #74's
finding and the reason the prefix list exists.

The existing mutation on that prefix reddens TestEveryRouteFallsUnderADeclaredPrefix,
which is a statement about the route table. This adds the one that reddens the
behaviour a client sees, and it bites: seven mutations, all of them.

Assisted-by: Claude Code (claude-opus-5)
@stephrobert
stephrobert merged commit 4d34fb4 into main Sep 15, 2026
31 checks passed
@stephrobert
stephrobert deleted the fix/a-stale-comment-about-the-gateway-404 branch September 15, 2026 08:10
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant