This changed the node pruning logic and moved test/functional/feature_blockfilterindex_prune.py
to test/functional/feature_index_prune.py.
Please verify that:
1. I migrated the test correctly
2. The magic numbers in the test look fine
With respect to #2: I believe the magic numbers are wrong. I previously had to tweak them heavily
in commit 1278b31. I don't think I did it correctly then, and so I don't believe them to be correct now.
To summarize what this tweaking was: I changed the magic numbers in the test to work properly,
but I suspect that in changing them, I may have nullified what the test was testing.
It's very possible that the reason the test was failing was because of an underlying bug with the pruning
in elements which we have to fix, rather than just being an issue with the test itself.
Ideally, this new test (added in last commit) should work in Elements, but I couldn't get
it working. Please look into if you can get this test working. Perhaps we need to set
g_con_elementsmode to False, or something similar.
The first transaction was failing with bad-txns-inputs-missingorspent
because the "input" selected with fundrawtransaction was the
anyonecanspend initialfreecoins.
Fixed by first spending the initialfreecoins to "real" outputs.
The first transaction from create_taproot_utxo was failing with
bad-txns-inputs-missingorspent because the "input" selected with
fundrawtransaction was the anyonecanspend initialfreecoins.
Fixed by first spending the initialfreecoins to "real" outputs.
Adds a functional test to cover the issue uncovered in #1259, where
calling fundrawtransaction with many non-policy inputs and no policy
recipients results in an assertion failure and a crash.
Fixed in #1258.
(cherry picked from commit a8b0ed6f96)
Fixes https://github.com/ElementsProject/elements/issues/1263
When not running in Elements mode, GetValidFedpegScripts returns an
empty vector. Subsequently `calculate_contract` is called with
`fedpegscripts.front().second` which is a null reference. This reference
is passed into `MatchLiquidWatchman` which tries to dereference the null
reference and causes a segfault.
This fix throws an exception in `getpeginaddress` if fedpegscripts is
empty, returning an error to the RPC caller. It also adds a functional
regression test.
In the "macOS 13 native arm64 [gui, sqlite only]" CI task, the
test_runner defaults to descriptor wallets, causing some Elements
specific tests to fail.
This commit explicitly runs those tests with `--legacy-wallet`.
