Skip to content

Fix estimate fee#1

Merged
freddiecoleman merged 2 commits into
Spacetime-Technology:mainfrom
Jsewill:fix-estimate-fee
May 21, 2026
Merged

Fix estimate fee#1
freddiecoleman merged 2 commits into
Spacetime-Technology:mainfrom
Jsewill:fix-estimate-fee

Conversation

@Jsewill
Copy link
Copy Markdown
Contributor

@Jsewill Jsewill commented May 20, 2026

(claude and) I found a bug in get_fee_estimate, while I was playing around with it, looking at making this work with a local full node, and experimenting with a full rust port.

Tool calls didn't seem to work except in a single scenario. As claude determined:
image

This pull request is meant to fix that, but you may have a better way to do it.

coinset's get_fee_estimate accepts only `cost` + `target_times` and rejects
`spend_type`/`spend_count`. estimate_fee forwarded those, so a bare call (or one
with `spend_type`) errored with "missing field `cost`" / "unknown field
`spend_type`" — only an explicit `cost` ever worked.

Resolve to a `cost` and send only `cost` + `target_times`: the explicit `cost`
if given, else the `spend_type`'s approximate cost (rough multiples of one
standard spend) or a standard spend, scaled by `spend_count`. The spend_type /
spend_count hints stay and are mapped to a cost client-side (coinset's own error
says "use cost instead"). Param/tool descriptions are unchanged.

Also fixes the unit test, which had mocked get_fee_estimate and asserted the
broken no-cost payload — so it stayed green while the real call failed.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
@Jsewill Jsewill force-pushed the fix-estimate-fee branch from e027c86 to eec75d5 Compare May 20, 2026 23:28
@freddiecoleman
Copy link
Copy Markdown
Member

Thanks for the bug report and contribution!

chia.net's Jan-2024 fees post puts offer-take cost at 80-150M. The
previous 44M (4x) under-shoots by ~2-3x, and because coinset ranks by
cost-per-fee, an under-estimated cost yields an under-estimated fee for
this spend_type — the user's transaction would sit longer than the
target_times they asked for. 9x (99M) sits in the middle of the
published range and matches the table's "multiples of one standard
spend" style.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
@freddiecoleman freddiecoleman merged commit a2a39db into Spacetime-Technology:main May 21, 2026
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.

2 participants