The analysis was wrongly displayed. For instance it could display a
fingerprint as having both signed and not signed yet.
Clean up the code and fix the analysis.
bdd16c73bcae64e17931bd62d70f4d85b9a6af9a installer: start new bitcoind download if not finished (jp1ac4)
Pull request description:
In case the user clicks on "Previous" before the bitcoind download has finished, this PR will clear the incomplete download (or a download that completed with errors) from `InternalBitcoindStep`.
If the user then returns to this step, either a new download will start or the finished download will be available still. This fixes a bug following #746 for users that already have bitcoind v25.0 installed. If they click "Previous" while downloading 25.1, then upon returning to this step, the existing 25.0 bitcoind will be started.
ACKs for top commit:
darosior:
ACK bdd16c73bcae64e17931bd62d70f4d85b9a6af9a -- code makes sense to me and Kevin tested it fixed (one of) the bug he reported.
Tree-SHA512: d1f39b1ee2b1f6fa15533f6c7880dc0e3edc9e7052be8ff315d77232eab9e74f8178252979c14cd8ccac8f164176cbe03a18faf523c969d841d11d6721b18fe1
9d9c4e57c2de0ae6e61bdf3d402dbfc172ed5ae3 doc: mention the bitbox too is supported now! (Antoine Poinsot)
Pull request description:
ACKs for top commit:
darosior:
self-ACK 9d9c4e57c2de0ae6e61bdf3d402dbfc172ed5ae3 -- trivial
Tree-SHA512: 09b3c38ec64014c0e961b5396b3ea70ec8d62cf40f5c61936a6e00344c26f5db44da55a0d29564e96ba953261119836b405a172708993e338c11b022a05b3fd0
In case user clicks on Previous before bitcoind download has
finished, this will clear the incomplete download from
`InternalBitcoindStep` so that a new download will start if the
user returns to this step.
d0065898f70fbe14be7d269a686110e3bbbe7af8 loader: don't read bitcoind's stdout to avoid a deadlock (edouardparis)
Pull request description:
"Forward port" of #770.
It was possible to get into a state where we were waiting for lianad to start, which was waiting on a response from bitcoind which in turn was waiting on the GUI to make room in its piped stdout buffer. Deadlock.
Instead of getting its logs from stdout, read its debug.log file.
ACKs for top commit:
edouardparis:
ACK d0065898f70fbe14be7d269a686110e3bbbe7af8
Tree-SHA512: 0c66190ba87b56b883d15463ad24e152e43b33bf235ba8fb2cab6978c6eace1fdf3b6280377548a92248b18b0fa05bd4ad698cb65beac92bb5f11bc8a4662052
a14f3afa2aab52f686c2bc0a445ca359282922ae installer: do not reload last step on next message (edouardparis)
Pull request description:
If a user clicked very fast a lot on the Next button to trigger multiple Next messages, then the last
step of the install reloads multiple time.
This commit introduce a check that ignore next message if the step is the last one.
ACKs for top commit:
darosior:
utACK a14f3afa2aab52f686c2bc0a445ca359282922ae -- code makes sense and the patch was tested by Kevin
Tree-SHA512: a0afb3eb0793e7d294bdf1d355816ea175f3652f0ce3b018150714ce22425a4ec1f8795173ac241cc43ad2d6dadb3668f0e0d7b720804cc83ad54cd99d9a27af
It was possible to get into a state where we were waiting for lianad to
start, which was waiting on a response from bitcoind which in turn was
waiting on the GUI to make room in its piped stdout buffer. Deadlock.
Instead of getting its logs from stdout, read its debug.log file.
Co-Authored-By: Antoine Poinsot <darosior@protonmail.com>
If a user clicked very fast a lot on the Next button
to trigger multiple Next messages, then the last
step of the install reloads multiple time.
This commit introduce a check that ignore next message
if the step is the last one.
d56a9a8a40671ad0ab0f4d8ccb8c6fa861d40429 fix single payment output label (edouardparis)
Pull request description:
When a transaction has only one payment,
then its txid label is attached to the payment label. While modifying the label of an output it did not change the whole transaction label, which is a bug.
backport of the fix for 3.x in #768
ACKs for top commit:
edouardparis:
Self-ACK d56a9a8a40671ad0ab0f4d8ccb8c6fa861d40429
Tree-SHA512: 0c2bca6f2a5527c37653c3644907926f6dacd01f826a2538a78c898fda407e49dceb40614ef8389ae7663228da1c1121decf4be21dc8e0f9a44fd788b3c76972
789f952433d7ccc0f09924359bd87b83b0bdcf55 fix gui: update labels if user signed an unsaved tx (edouardparis)
Pull request description:
It is a fix from #760
ACKs for top commit:
edouardparis:
Self-ACK 789f952433d7ccc0f09924359bd87b83b0bdcf55
Tree-SHA512: 3424577ba6d3215a3b7503a0861997ace0dbac0a82b0e43af5bc5fb5262450ab5074eaa54ce6a01021237b85dfb163d4eb436916d17f52b9dfd1a07cf7be269c
When a transaction has only one payment,
then its txid label is attached to the payment label.
While modifying the label of an output it did not change the
whole transaction label, which is a bug.
9a4b665fa1a471a7a6c7e0dea09065f9f01ec35d fix wallet name in settings during install (edouardparis)
Pull request description:
DEFAULT_WALLET_NAME aka simple Liana was still used during creation of wallet settings during install
this should have been noticed if I did test the wallet creation process in detail and if I removed the public exposition of the const variable. SHAME
ACKs for top commit:
edouardparis:
Self-ACK 9a4b665fa1a471a7a6c7e0dea09065f9f01ec35d
Tree-SHA512: 3afd98918a7ddcd10e702d0aed66a0645bea734fde61febe021f07c67f5b42a28018e1556e052f03e413db9c75686f121091a7ab871c7c3533d62b7202761625
5f0abbcc877fe0c31d11b21a2ce1ce7146dbbf38 Cargo: drop two keywords (Antoine Poinsot)
Pull request description:
We had too many keywords to be able to publish on crates.io
ACKs for top commit:
darosior:
self-ACK 5f0abbcc877fe0c31d11b21a2ce1ce7146dbbf38 -- trivial
Tree-SHA512: daf57c3f036ff44c164fa80eb8a88afdb799a63d3a59e19423eacba579300bc76ec0a1168a9ae9260484b754b0695ae98da4b4ed7a0cf3eff885010d15fcec04
ea4c4462406f5a1860140ca92bb7fca34e1322e0 descriptor: add a helper which guesses which paths to prune from a PSBT (Antoine Poinsot)
d03e603abbd588d97da81cf0e51d28cbef82f94d descriptors: add a helper to prune BIP32 derivation paths (Antoine Poinsot)
Pull request description:
Based on #749, this provides clients of the library with a way to prune all the BIP32 derivation paths in a PSBT for keys which are not part of the given spending path.
ACKs for top commit:
edouardparis:
utACK ea4c4462406f5a1860140ca92bb7fca34e1322e0
Tree-SHA512: f9f21e12e562a047693066e7b54c2601de01d79071ac40981406a04cf74bfc2da4111b5ea251e3a399a0df95ff6f873da7a7cbb112251100c81bfba171b0a21a
1cbbb8c0befe3c168fbbd34129a2f4f2a8c8726f wallet: load name from settings (edouardparis)
Pull request description:
If an wallet settings already exists, the wallet name attached to the hardware wallet credentials must be use, instead of the new generated one.
close#745
ACKs for top commit:
darosior:
ACK 1cbbb8c0befe3c168fbbd34129a2f4f2a8c8726f -- tested locally and discussed it with Edouard. I can't think of another place where we'd like to use the old wallet name.
Tree-SHA512: a06e64f4ada4fbadb6e132be740b480055113e733d658f4694cfcad15bb2901e6626af211ba5df95f4e08d9cb47080a96e97e29bf1c18e094dfbde6364408ce2
17ca01322e7badd37fd4120e0aeec8d834d0d6df Revert "Merge #722: Only include BIP32 derivations for a single spending path when creating PSBTs" (Antoine Poinsot)
6f5b053ea5af4737885a25573c82d627c299236e Revert "Merge #742: commands: don't add derivation paths for keys from different path but same signer" (Antoine Poinsot)
Pull request description:
#722 pruned the BIP32 derivation paths in PSBT inputs for other spending paths than the one this PSBT was created for. This was to make it possible to use them with the Bitbox (see #722 OP for more details).
These were merged without testing at all whether it would break signing with a Ledger Nano. Turns out it did. Our bad. Thanks to Pythcoiner for finding this bug.
A follow-up PR will add a helper to prune the derivation separately so that it can only be applied to some signers and not all of them.
ACKs for top commit:
edouardparis:
utACK 17ca01322e7badd37fd4120e0aeec8d834d0d6df
Tree-SHA512: 99764fc5c68bc66e8415d72bb9963822ea05f8447c1ec71b50db992f3bf456370e00ae59055c939df1817c0031cdfbbfddbcdae9c448790fdcd67df10589b43d
d8d01df6070bdd470b46de90dfcae4a85f704812 contrib: correct minimum libc version in Debian package (Antoine Poinsot)
Pull request description:
It was 2.34, it should be 2.33.
ACKs for top commit:
darosior:
self ACK d8d01df6070bdd470b46de90dfcae4a85f704812 -- trivial
Tree-SHA512: 0ea2fd6b987636a632dd82e51252f2eb11b194caf6b26d97f897fde7b14fd5da6354a6c02a733d3feaf666363eecfaf4974b64af2deb32247ce360967e8b851a
The call to getblockheader may fail, for instance when bitcoind is
rolling forward its blocks after a crash. It's a pretty edgy case but
instead of crashing restart over and over again.
Edouard tells me it would help a lot as it would avoid having to keep
state in the GUI: otherwise it would need to know in a signing sessions
with a Bitbox whether it's signing a spend or recovery PSBT (and for
which recovery path).
fbec5e9218b24190c8f0ae1e1a09baf808ac70fa gui: bump managed bitcoind version to 25.1 (jp1ac4)
1808f8f1de5efbb402161829c4bab25a985e733a gui: allow to bump managed bitcoind version (jp1ac4)
Pull request description:
This PR bumps the managed bitcoind version for new Liana installations to 25.1.
The GUI will look for the most recent managed bitcoind version available when starting bitcoind.
Once a new managed bitcoind version is installed, all wallets using managed bitcoind will use this new version, even if a wallet on a particular network was installed with a previous bitcoind version.
The approach taken here may change as part of #690.
ACKs for top commit:
darosior:
ACK fbec5e9218b24190c8f0ae1e1a09baf808ac70fa -- neat PR.
Tree-SHA512: e6067e00e6c55bc112b0d9b873788de7edd2d3157e58ea1bbcbf512d2f2d621629d32c36010e7602e08e2a17f053a8f5690caecab7f86119b6a7b381f7b11df1
This reverts commit 71056982636b408485ab24dab6628a555a6e7924, reversing
changes made to 03c37bd378f4f6bf11d90b224ed1db74b3596eaf.
This reverts PR #722. It turns out the Ledger Bitcoin app needs the
BIP32 derivation for all the keys in the Script, not only for the
spending path used. Therefore always create PSBT with all the BIP32
derivations. We'll add a way to prune them for talking to the Bitbox in
a future commit.
This reverts commit ec0c2426aa5fa6cee2efabd3ee6f175b41c35f64, reversing
changes made to 26d750d09c84734f56c2dc18cb332a232e24fb6d.
This reverts the fixes to the pruning of BIP32 derivation paths when
creating a PSBT, in preparation of reverting the merge of this feature
altogether. This is because always creating PSBT with only the BIP32
derivations for a single path broke the Ledger support.
This makes it possible to bump the managed bitcoind version used
when installing a new Liana wallet.
The GUI will look for the most recent managed bitcoind version
available when starting bitcoind so that existing wallets
with a previous managed bitcoind version will not be affected.
4be74ad496692d371adffcc1b3c868ef6a01cb52 refac hw module and add bitbox support (edouard)
Pull request description:
- Add bitbox02 as possible signer to liana-gui
- Introduce global_settings.json in datadir to keep the bitbox noise config.
ACKs for top commit:
edouardparis:
Self-ACK 4be74ad496692d371adffcc1b3c868ef6a01cb52
Tree-SHA512: e0615e1903baa7faecc25db717f49ed71c23a6afcec607c41ac2045cfb2cfa9a9aa3d3452cc952aeaf72a735d670c56dbf19b239c8d665466a2ecb09fbe222c5
2f2975c2790e83c592cd0ff594d3e8ce8a7c75d9 coinbase transaction outputs may not have valid scripts (edouard)
Pull request description:
close#738
liana-gui v2 also tried to display addresses from transaction outputs to user when clicking on transaction detail.
ACKs for top commit:
edouardparis:
Self-ACK 2f2975c2790e83c592cd0ff594d3e8ce8a7c75d9
Tree-SHA512: 7f5fc374baec773fef848cd104b8acbd34ce19a04f9614d0c446a93ee8b6b976acd98d7494031651a614bef38c0913217865f2086bf19a018ce17f57ef362c57
2e1c54491e96fda79b3831dc0c079c59250fabe0 qa: test der paths in PSBT when desc has duplicate signer (Antoine Poinsot)
8d213d5e31dd47f0c179475e32f5eb86c3773770 qa: correct create_spend functional test (Antoine Poinsot)
cf33228b0d25e79ec8fa855aaac6ebfa5f10da12 qa: don't assume desc xpubs' der path length in finalizer (Antoine Poinsot)
24edaecbdcfc0b6b53ac3355ad22991a8b5de018 commands: don't add der paths for keys from diff path but same signer (Antoine Poinsot)
Pull request description:
When creating a PSBT, we were checking whether the key was for this path by checking its origin. This check would return false positive for keys from other paths but same signer (which shares the same fingerprint).
Instead, check the entire origin for each key to make sure it's actually the one used in the path we are interested about.
Thanks to Edouard Paris for finding this bug.
ACKs for top commit:
darosior:
self-ACK 2e1c54491e96fda79b3831dc0c079c59250fabe0 - only made the test more robust since Edouard's ACK
Tree-SHA512: 3bbd0db9e1be59318f9e0af7b6ebf2d91d919e4946acb3dd02822cd6c0e2b9829a963ff1ecd9b0b0ab780bd46b4f0d120ab80b2f55ee84fbcf848231727b3cac
