Now, we can use object storage (MinIO, S3, R2 or anything you like) for synchronising! Moreover, despite that, we can use all the features as if we were using CouchDB.
Note: As this is a pretty experimental feature, hence we have some limitations.
- This is built on the append-only architecture. It will not shrink used storage if we do not perform a rebuild.
- A bit fragile. However, our version x.yy.0 is always so.
- When the first synchronisation, the entire history to date is transferred. For this reason, it is preferable to do this under the WiFi network.
- Do not worry, from the second synchronisation, we always transfer only differences.
I hope this feature empowers users to maintain independence and self-host their data, offering an alternative for those who prefer to manage their own storage solutions and avoid being stuck on the right side of a sudden change in business model.
Of course, I use Self-hosted MinIO for testing and recommend this. It is for the same reason as using CouchDB. -- open, controllable, auditable and indeed already audited by numerous eyes.
Let me write one more acknowledgement.
I have a lot of respect for that plugin, even though it is sometimes treated as if it is a competitor, remotely-save. I think it is a great architecture that embodies a different approach to my approach of recreating history. This time, with all due respect, I have used some of its code as a reference.
Hooray for open source, and generous licences, and the sharing of knowledge by experts.
- The hodgepodge inside the `Hatch` pane has been sorted into more explicit categorised panes.
- Now we have new panes for:
-`Selector`
-`Advanced`
-`Power users`
-`Patches (Edge case)`
- Applying the settings will now be more informative.
- The header bar will be shown for applying the settings which needs a database rebuild.
- Applying methods are now more clearly navigated.
- Definitely, drastic change. I hope this will be more user-friendly. However, if you notice any issues, please let me know. I hope that nothing missed.
- New features:
- Word-segmented chunk building on users language
- Chunks can now be built with word-segmented data, enhancing efficiency for markdown files which contains the multiple sentences in a single line.
- This feature is enabled by default through `Use Segmented-splitter`.
- (Default: Disabled, Please be relived, I have learnt).
- Fixed:
- Sending chunks on `Send chunk in bulk` are now buffered to avoid the out-of-memory error.
-`Send chunk in bulk` is back to default disabled. (Sorry, not applied to the migrated users; I did not think we should deepen the wound any further "automatically").
- Merging conflicts of JSON files are now works fine even if it contains `null`.
- Development:
- Implemented the logic for automatically generating the stub of document for the setting dialogue.
- 0.23.22:
- Fixed:
- Case-insensitive file handling
- Full-lower-case files are no longer created during database checking.
- Bulk chunk transfer
- The default value will automatically adjust to an acceptable size when using IBM Cloudant.
- 0.23.21:
- New Features:
- Case-insensitive file handling
- Files can now be handled case-insensitively.
- This behaviour can be modified in the settings under `Handle files as Case-Sensitive` (Default: Prompt, Enabled for previous behaviour).
- Improved chunk revision fixing
- Revisions for chunks can now be fixed for faster chunk creation.
- This can be adjusted in the settings under `Compute revisions for chunks` (Default: Prompt, Enabled for previous behaviour).
- Bulk chunk transfer
- Chunks can now be transferred in bulk during uploads.
- This feature is enabled by default through `Send chunks in bulk`.
- Creation of missing chunks without
- Missing chunks can be created without storing notes, enhancing efficiency for first synchronisation or after prolonged periods without synchronisation.
- Improvements:
- File status scanning on the startup
- Quite significant performance improvements.
- No more missing scans of some files.
- Status in editor enhancements
- Significant performance improvements in the status display within the editor.
- Notifications for files that will not be synchronised will now be properly communicated.
- Encryption and Decryption
- These processes are now performed in background threads to ensure fast and stable transfers.
- Verify and repair all files
- Got faster through parallel checking.
- Migration on update
- Migration messages and wizards have become more helpful.
- Behavioural changes:
- Chunk size adjustments
- Large chunks will no longer be created for older, stable files, addressing storage consumption issues.
- Flag file automation
- Confirmation will be shown and we can cancel it.
- Fixed:
- Database File Scanning
- All files in the database will now be enumerated correctly.
- Miscellaneous
- Dependency updated.
- Now, tree shaking is left to terser, from esbuild.
- 0.23.20:
- Fixed:
- Customisation Sync now checks the difference while storing or applying the configuration.
- No longer storing the same configuration multiple times.
- Time difference in the dialogue has been fixed.
- Remote Storage Limit Notification dialogue has been fixed, now the chosen value is saved.
- Improved:
- The Enlarging button on the enlarging threshold dialogue now displays the new value.
- Per-file-saved customization sync has been shipped.
- We can synchronise plug-igs etc., more smoothly.
- Default: disabled. We need a small migration when enabling this. And all devices should be updated to v0.23.18. Once we enabled this, we lost compatibility with old versions.
- Customisation sync has got beta3.
- We can set `Flag` to each item to select the newest, automatically.
- This configuration is per device.
- Improved:
- Start-up speed has been improved.
- Fixed:
- On the customisation sync dialogue, buttons are kept within the screen.
- No more unnecessary entries on `data.json` for customisation sync.
- Selections are no longer lost while updating customisation items.
- Tidied on source codes:
- Many typos have been fixed.
- Some unnecessary type casting removed.
- 0.23.17:
- Improved:
- Overall performance has been improved by using PouchDB 9.0.0.
- Configuration mismatch detection is refined. We can resolve mismatches more smoothly and naturally.
More detail is on `troubleshooting.md` on the repository.
- Fixed:
- Customisation Sync will be disabled when a corrupted configuration is detected.
Therefore, the Device Name can be changed even in the event of a configuration mismatch.
- New feature:
- We can get a notification about the storage usage of the remote database.
- Default: We will be asked.
- If the remote storage usage approaches the configured value, we will be asked whether we want to Rebuild or increase the limit.
- The setting dialogue has been refined. Very controllable, clearly displayed disabled items, and ready to i18n.
- Fixed:
- Many memory leaks have been rescued.
- Chunk caches now work well.
- Many trivial but potential bugs are fixed.
- No longer error messages will be shown on retrieving checkpoint or server information.
- Now we can check and correct tweak mismatch during the setup
- Improved:
- Customisation synchronisation has got more smoother.
- Tidied
- Practically unused functions have been removed or are being prepared for removal.
- Many of the type-errors and lint errors have been corrected.
- Unused files have been removed.
- Note:
- From this version, some test files have been included. However, they are not enabled and released in the release build.
- To try them, please run Self-hosted LiveSync in the dev build.
- 0.23.7
- Fixed:
- No longer missing tasks which have queued as the same key (e.g., for the same operation to the same file).
- This occurs, for example, with hidden files that have been changed multiple times in a very short period of time, such as `appearance.json`. Thanks for the report!
- Some trivial issues have been fixed.
- New feature:
- Reloading Obsidian can be scheduled until that file and database operations are stable.
- Now the remote chunks could be decrypted even if we are using `Incubate chunks in Document`. (The note of 0.23.6 has been fixed).
- Chunk retrieving with `Incubate chunks in document` got more efficiently.
- No longer task processor misses the completed tasks.
- Replication is no longer started automatically during changes in window visibility (e.g., task switching on the desktop) when off-focused.
- 0.23.5:
- New feature:
- Now we can check configuration mismatching between clients before synchronisation.
- Default: enabled / Preferred: enabled / We can disable this by the `Do not check configuration mismatch before replication` toggle in the `Hatch` pane.
- It detects configuration mismatches and prevents synchronisation failures and wasted storage.
- Now we can perform remote database compaction from the `Maintenance` pane.
- Fixed:
- We can detect the bucket could not be reachable.
- Note:
- Known inexplicable behaviour: Recently, (Maybe while enabling `Incubate chunks in Document` and `Fetch chunks on demand` or some more toggles), our customisation sync data is sometimes corrupted. It will be addressed by the next release.
- No longer experimental configuration is shown on the Minimal Setup.
- New feature:
- We can now use `Incubate Chunks in Document` to reduce non-well-formed chunks.
- Default: disabled / Preferred: enabled in all devices.
- When we enabled this toggle, newly created chunks are temporarily kept within the document, and graduated to become independent chunks once stabilised.
- The [design document](https://github.com/vrtmrz/obsidian-livesync/blob/3925052f9290b3579e45a4b716b3679c833d8ca0/docs/design_docs_of_keep_newborn_chunks.md) has been also available..
- 0.23.3
- Fixed: No longer unwanted `\f` in journal sync.
- 0.23.2
- Sorry for all the fixes to experimental features. (These things were also critical for dogfooding). The next release would be the main fixes! Thank you for your patience and understanding!
- Fixed:
- Journal Sync will not hang up during big replication, especially the initial one.
- All changes which have been replicated while rebuilding will not be postponed (Previous behaviour).
- Improved:
- Now Journal Sync works efficiently in download and parse, or pack and upload.
- Less server storage and faster packing/unpacking usage by the new chunk format.
- 0.23.1
- Fixed:
- Now journal synchronisation considers untransferred each from sent and received.
- Journal sync now handles retrying.
- Journal synchronisation no longer considers the synchronisation of chunks as revision updates (Simply ignored).
- Journal sync now splits the journal pack to prevent mobile device rebooting.
- Maintenance menus which had been on the command palette are now back in the maintain pane on the setting dialogue.
- Improved:
- Now all changes which have been replicated while rebuilding will be postponed.
A few years passed since Self-hosted LiveSync was born, and our codebase had been very complicated. This could be patient now, but it should be a tremendous hurt.
Therefore at v0.22.0, for future maintainability, I refined task scheduling logic totally.
Of course, I think this would be our suffering in some cases. However, I would love to ask you for your cooperation and contribution.
Sorry for being absent so much long. And thank you for your patience!
Note: we got a very performance improvement.
Note at 0.22.2: **Now, to rescue mobile devices, Maximum file size is set to 50 by default**. Please configure the limit as you need. If you do not want to limit the sizes, set zero manually, please.
- No longer data corrupting due to false BASE64 detections.
- Improved:
- A bit more efficient in Automatic data compression.
- 0.22.18
- New feature (Very Experimental):
- Now we can use `Automatic data compression` to reduce amount of traffic and the usage of remote database.
- Please make sure all devices are updated to v0.22.18 before trying this feature.
- If you are using some other utilities which connected to your vault, please make sure that they have compatibilities.
- Note: Setting `File Compression` on the remote database works for shrink the size of remote database. Please refer the [Doc](https://docs.couchdb.org/en/stable/config/couchdb.html#couchdb/file_compression).
- 0.22.17:
- Fixed:
- Error handling on booting now works fine.
- Replication is now started automatically in LiveSync mode.
- Batch database update is now disabled in LiveSync mode.
- No longer automatically reconnection while off-focused.
- Status saves are thinned out.
- Now Self-hosted LiveSync waits for all files between the local database and storage to be surely checked.
- Improved:
- The job scheduler is now more robust and stable.
- The status indicator no longer flickers and keeps zero for a while.
- No longer meaningless frequent updates of status indicators.
- Now we can configure regular expression filters in handy UI. Thank you so much, @eth-p!
-`Fetch` or `Rebuild everything` is now more safely performed.
- Minor things
- Some utility function has been added.
- Customisation sync now less wrong messages.
- Digging the weeds for eradication of type errors.
- 0.22.16:
- Fixed:
- Fixed the issue that binary files were sometimes corrupted.
- Fixed customisation sync data could be corrupted.
- Improved:
- Now the remote database costs lower memory.
- This release requires a brief wait on the first synchronisation, to track the latest changeset again.
- Description added for the `Device name`.
- Refactored:
- Many type-errors have been resolved.
- Obsolete file has been deleted.
- 0.22.15:
- Improved:
- Faster start-up by removing too many logs which indicates normality
- By streamlined scanning of customised synchronisation extra phases have been deleted.
The E2EE encryption V2 format has been reverted. That was probably the cause of the glitch.
Instead, to maintain efficiency, files are treated with Blob until just before saving. Along with this, the old-fashioned encryption format has also been discontinued.
There are both forward and backwards compatibilities, with recent versions. However, unfortunately, we lost compatibility with filesystem-livesync or some.
It will be addressed soon. Please be patient if you are using filesystem-livesync with E2EE.
- 0.21.5
- Improved:
- Now all revisions will be shown only its first a few letters.
- Now ID of the documents is shown in the log with the first 8 letters.
- Fixed:
- Check before modifying files has been implemented.
- Content change detection has been improved.
- 0.21.4
- This release had been skipped.
- 0.21.3
- Implemented:
- Now we can use SHA1 for hash function as fallback.
- 0.21.2
- IMPORTANT NOTICE: **0.21.1 CONTAINS A BUG WHILE REBUILDING THE DATABASE. IF YOU HAVE BEEN REBUILT, PLEASE MAKE SURE THAT ALL FILES ARE SANE.**
- This has been fixed in this version.
- Fixed:
- No longer files are broken while rebuilding.
- Now, Large binary files can be written correctly on a mobile platform.
- Any decoding errors now make zero-byte files.
- Modified:
- All files are processed sequentially for each.
- 0.21.1
- Fixed:
- No more infinity loops on larger files.
- Show message on decode error.
- Refactored:
- Fixed to avoid obsolete global variables.
- 0.21.0
- Changes and performance improvements:
- Now the saving files are processed by Blob.
- The V2-Format has been reverted.
- New encoding format has been enabled in default.
- WARNING: Since this version, the compatibilities with older Filesystem LiveSync have been lost.
At 0.20.0, Self-hosted LiveSync has changed the binary file format and encrypting format, for efficient synchronisation.
The dialogue will be shown and asks us to decide whether to keep v1 or use v2. Once we have enabled v2, all subsequent edits will be saved in v2. Therefore, devices running 0.19 or below cannot understand this and they might say that decryption error. Please update all devices.
Then we will have an impressive performance.
Of course, these are very impactful changes. If you have any questions or troubled things, please feel free to open an issue and mention me.
Note: if you want to roll it back to v1, please enable `Use binary and encryption version 1` on the `Hatch` pane and perform the `rebuild everything` once.
Extra but notable information:
This format change gives us the ability to detect some `marks` in the binary files as same as text files. Therefore, we can split binary files and some specific sort of them (i.e., PDF files) at the specific character. It means that editing the middle of files could be detected with marks.
Now only a few chunks are transferred, even if we add a comment to the PDF or put new files into the ZIP archives.
- 0.20.7
- Fixed
- To better replication, path obfuscation is now deterministic even if with E2EE.
Note: Compatible with previous database without any conversion. Only new files will be obfuscated in deterministic.
- 0.20.6
- Fixed
- Now empty file could be decoded.
- Local files are no longer pre-saved before fetching from a remote database.
- No longer deadlock while applying customisation sync.
- Configuration with multiple files is now able to be applied correctly.
- Deleting folder propagation now works without enabling the use of a trash bin.
- 0.20.5
- Fixed
- Now the files which having digit or character prefixes in the path will not be ignored.
- 0.20.4
- Fixed
- The text-input-dialogue is no longer broken.
- Finally, we can use the Setup URI again on mobile.
- 0.20.3
- New feature:
- We can launch Customization sync from the Ribbon if we enabled it.
- Fixed:
- Setup URI is now back to the previous spec; be encrypted by V1.
- It may avoid the trouble with iOS 17.
- The Settings dialogue is now registered at the beginning of the start-up process.
- We can change the configuration even though LiveSync could not be launched in normal.
- Improved:
- Enumerating documents has been faster.
- 0.20.2
- New feature:
- We can delete all data of customization sync from the `Delete all customization sync data` on the `Hatch` pane.
- Fixed:
- Prevent keep restarting on iOS by yielding microtasks.
- 0.20.1
- Fixed:
- No more UI freezing and keep restarting on iOS.
- Diff of Non-markdown documents are now shown correctly.
- Improved:
- Performance has been a bit improved.
- Customization sync has gotten faster.
- However, We lost forward compatibility again (only for this feature). Please update all devices.
- Misc
- Terser configuration has been more aggressive.
- 0.20.0
- Improved:
- A New binary file handling implemented
- A new encrypted format has been implemented
- Now the chunk sizes will be adjusted for efficient sync
- Fixed:
- levels of exception in some logs have been fixed
Since `Plugin and their settings` have been broken, so I tried to fix it, not just fix it, but fix it the way it should be.
Now, we have `Customization sync`.
It is a real shame that the compatibility between these features has been broken. However, this new feature is surely useful and I believe that worth getting over the pain.
We can use the new feature with the same configuration. Only the menu on the command palette has been changed. The dialog can be opened by `Show customization sync dialog`.
I hope you will give it a try.
#### Minors
- 0.19.1
- Fixed: Fixed hidden file handling on Linux
- Improved: Now customization sync works more smoothly.
- 0.19.2
- Fixed:
- Fixed garbage collection error while unreferenced chunks exist many.
- Fixed filename validation on Linux.
- Improved:
- Showing status is now thinned for performance.
- Enhance caching while collecting chunks.
- 0.19.3
- Improved:
- Now replication will be paced by collecting chunks. If synchronisation has been deadlocked, please enable `Do not pace synchronization` once.
- 0.19.4
- Improved:
- Reduced remote database checking to improve speed and reduce bandwidth.
- Fixed:
- Chunks which previously misinterpreted are now interpreted correctly.
- No more missing chunks which not be found forever, except if it has been actually missing.
- Deleted file detection on hidden file synchronising now works fine.
- Now the Customisation sync is surely quiet while it has been disabled.
- 0.19.5
- Fixed:
- Now hidden file synchronisation would not be hanged, even if so many files exist.
- Improved:
- Customisation sync works more smoothly.
- Note: Concurrent processing has been rollbacked into the original implementation. As a result, the total number of processes is no longer shown next to the hourglass icon. However, only the processes that are running concurrently are shown.
- 0.19.6
- Fixed:
- Logging has been tweaked.
- No more too many planes and rockets.
- The batch database update now surely only works in non-live mode.
- Internal things:
- Some frameworks has been upgraded.
- Import declaration has been fixed.
- Improved:
- The plug-in now asks to enable a new adaptor, when rebuilding, if it is not enabled yet.
- The setting dialogue refined.
- Configurations for compatibilities have been moved under the hatch.
- Made it clear that disabled is the default.
- Ambiguous names configuration have been renamed.
- Items that have no meaning in the settings are no longer displayed.
#### Now, paths of files in the database can now be obfuscated. (Experimental Feature)
At before v0.18.0, Self-hosted LiveSync used the path of files, to detect and resolve conflicts. In naive. The ID of the document stored in the CouchDB was naturally the filename.
However, it means a sort of lacking confidentiality. If the credentials of the database have been leaked, the attacker (or an innocent bystander) can read the path of files. So we could not use confidential things in the filename in some environments.
Since v0.18.0, they can be obfuscated. so it is no longer possible to decipher the path from the ID. Instead of that, it costs a bit CPU load than before, and the data structure has been changed a bit.
We can configure the `Path Obfuscation` in the `Remote database configuration` pane.
Note: **When changing this configuration, we need to rebuild both of the local and the remote databases**.
#### Minors
- 0.18.1
- Fixed:
- Some messages are fixed (Typo)
- File type detection now works fine!
- 0.18.2
- Improved:
- The setting pane has been refined.
- We can enable `hidden files sync` with several initial behaviours; `Merge`, `Fetch` remote, and `Overwrite` remote.
- No longer `Touch hidden files`.
- 0.18.3
- Fixed Pop-up is now correctly shown after hidden file synchronisation.
- 0.18.4
- Fixed:
-`Fetch` and `Rebuild database` will work more safely.
- Case-sensitive renaming now works fine.
Revoked the logic which was made at #130, however, looks fine now.
- 0.18.5
- Improved:
- Actions for maintaining databases moved to the `🎛️Maintain databases`.
- Clean-up of unreferenced chunks has been implemented on an **experimental**.
- This feature requires enabling `Use new adapter`.
- Be sure to fully all devices synchronised before perform it.
- After cleaning up the remote, all devices will be locked out. If we are sure had it be synchronised, we can perform only cleaning-up locally. If not, we have to perform `Fetch`.
- 0.18.6
- New features:
- Now remote database cleaning-up will be detected automatically.
- A solution selection dialogue will be shown if synchronisation is rejected after cleaning or rebuilding the remote database.
- During fetching or rebuilding, we can configure `Hidden file synchronisation` on the spot.
- It let us free from conflict resolution on initial synchronising.
- 0.17.0 has no surfaced changes but the design of saving chunks has been changed. They have compatibility but changing files after upgrading makes different chunks than before 0.16.x.
Please rebuild databases once if you have been worried about storage usage.
- Improved:
- Splitting markdown
- Saving chunks
- Changed:
- Chunk ID numbering rules
#### Minors
- 0.17.1
- Fixed: Now we can verify and repair the database.
- Refactored inside.
- 0.17.2
- New feature
- We can merge conflicted documents automatically if sensible.
- Fixed
- Writing to the storage will be pended while they have conflicts after replication.
- 0.17.3
- Now we supported canvas! And conflicted JSON files are also synchronised with merging its content if they are obvious.
- 0.17.4
- Canvases are now treated as a sort of plain text file. now we transfer only the metadata and chunks that have differences.
- 0.17.5 Now `read chunks online` had been fixed, and a new feature: `Use dynamic iteration count` to reduce the load on encryption/decryption.
Note: `Use dynamic iteration count` is not compatible with earlier versions.
- 0.17.6 Now our renamed/deleted files have been surely deleted again.
- 0.17.7
- Fixed:
- Fixed merging issues.
- Fixed button styling.
- Changed:
- Conflict checking on synchronising has been enabled for every note in default.
- 0.17.8
- Improved: Performance improved. Prebuilt PouchDB is no longer used.
- Fixed: Merging hidden files is also fixed.
- New Feature: Now we can synchronise automatically after merging conflicts.
- 0.17.9
- Fixed: Conflict merge of internal files is no longer broken.
- Improved: Smoother status display inside the editor.
- 0.17.10
- Fixed: Large file synchronising has been now addressed!
Note: When synchronising large files, we have to set `Chunk size` to lower than 50, disable `Read chunks online`, `Batch size` should be set 50-100, and `Batch limit` could be around 20.
- 0.17.11
- Fixed:
- Performance improvement
- Now `Chunk size` can be set to under one hundred.
- New feature:
- The number of transfers required before replication stabilises is now displayed.
- 0.17.12: Skipped.
- 0.17.13
- Fixed: Document history is now displayed again.
- Reorganised: Many files have been refactored.
- 0.17.14: Skipped.
- 0.17.15
- Improved:
- Confidential information has no longer stored in data.json as is.
- Synchronising progress has been shown in the notification.
- We can commit passphrases with a keyboard.
- Configuration which had not been saved yet is marked now.
- Now the filename is shown on the Conflict resolving dialog
- Now hidden files need not be scanned. Changes will be detected automatically.
- If you want it to back to its previous behaviour, please disable `Monitor changes to internal files`.
- Due to using an internal API, this feature may become unusable with a major update. If this happens, please disable this once.
#### Minors
- 0.16.1 Added missing log updates.
- 0.16.2 Fixed many problems caused by combinations of `Sync On Save` and the tracking logic that changed at 0.15.6.
- 0.16.3
- Fixed detection of IBM Cloudant (And if there are some issues, be fixed automatically).
- A configuration information reporting tool has been implemented.
- 0.16.4 Fixed detection failure. Please set the `Chunk size` again when using a self-hosted database.
- 0.16.5
- Fixed
- Conflict detection and merging now be able to treat deleted files.
- Logs while the boot-up sequence has been tidied up.
- Fixed incorrect log entries.
- New Feature
- The feature of automatically deleting old expired metadata has been implemented.
We can configure it in `Delete old metadata of deleted files on start-up` in the `General Settings` pane.
- 0.16.6
- Fixed
- Automatic (temporary) batch size adjustment has been restored to work correctly.
- Chunk splitting has been backed to the previous behaviour for saving them correctly.
- Improved
- Corrupted chunks will be detected automatically.
- Now on the case-insensitive system, `aaa.md` and `AAA.md` will be treated as the same file or path at applying changesets.
- 0.16.7 Nothing has been changed except toolsets, framework library, and as like them. Please inform me if something had been getting strange!
- 0.16.8 Now we can synchronise without `bad_request:invalid UTF-8 JSON` even while end-to-end encryption has been disabled.
Note:
Before 0.16.5, LiveSync had some issues making chunks. In this case, synchronisation had became been always failing after a corrupted one should be made. After 0.16.6, the corrupted chunk is automatically detected. Sorry for troubling you but please do `rebuild everything` when this plug-in notified so.
### 0.15.0
- Outdated configuration items have been removed.
- Setup wizard has been implemented!
I appreciate for reviewing and giving me advice @Pouhon158!
#### Minors
- 0.15.1 Missed the stylesheet.
- 0.15.2 The wizard has been improved and documented!
- 0.15.3 Fixed the issue about locking/unlocking remote database while rebuilding in the wizard.
- 0.15.4 Fixed issues about asynchronous processing (e.g., Conflict check or hidden file detection)
- 0.15.5 Add new features for setting Self-hosted LiveSync up more easier.
- 0.15.6 File tracking logic has been refined.
- 0.15.7 Fixed bug about renaming file.
- 0.15.8 Fixed bug about deleting empty directory, weird behaviour on boot-sequence on mobile devices.
- 0.15.9 Improved chunk retrieving, now chunks are retrieved in batch on continuous requests.
- 0.15.10 Fixed:
- The boot sequence has been corrected and now boots smoothly.
- Auto applying of batch save will be processed earlier than before.
- The metadata of the deleted files will be kept on the database by default. If you want to delete this as the previous version, please turn on `Delete metadata of deleted files.`. And, if you have upgraded from the older version, please ensure every device has been upgraded.
- Please turn on `Delete metadata of deleted files.` if you are using livesync-classroom or filesystem-livesync.
- We can see the history of deleted files.
-`Pick file to show` was renamed to `Pick a file to show.
- Files in the `Pick a file to show` are now ordered by their modified date descent.
- Update information became to be shown on the major upgrade.
#### Minors
- 0.13.1 Fixed on conflict resolution.
- 0.13.2 Fixed file deletion failures.
- 0.13.4
- Now, we can synchronise hidden files that conflicted on each devices.
- We can search for conflicting docs.
- Pending processes can now be run at any time.
- Performance improved on synchronising large numbers of files at once.
