- Introduced a new `validate` callback for `tag.define`
- Added `index.validateObjects` API to validate and transform objects before indexing.
- Updated relevant documentation and examples to reflect these changes.
- Introduced `currentPageMeta` method to encapsulate access to the current page's metadata.
- Updated references to page metadata throughout the codebase to use the new method, improving code readability and maintainability.
- Adjusted related components in the UI and linting logic to utilize the new method for consistency.
* Removed code complete from mini editor (didn't really work anyway, and
only use case was emoji)
* Changed meta picker to be more consistent with page picker
* Alt-space completion in page picker (completes "folder" to the top
match)
* Upgraded fuse library
* Removed the category support in the filterbox, it was buggy and hides
command names
* Tweaked fuse parameters for (hopefully) better matches in the page
picker (and everywhere where the filter box is used)
* Moved YAML library import to index plug (so no async syscalls anymore)
should speed things up
* Merged task and item indexing
* Merged tasks plug into index (questionable)
* taskstate is now no longer a tag, but needs to be configured via
configs
When the server returns HTTP 500 or other 5xx errors, the client
was caching the error HTML page (e.g., Nginx error pages) and
attempting to parse it as JSON on boot, causing a lockout with
the message 'Could not process config and no cached copy'.
This fix adds all standard 5xx server error codes to the
offlineStatusCodes list, ensuring the client enters offline mode
and uses cached data instead of caching error responses.
Before: Only 502, 503, 504, 530 triggered offline mode
After: All 5xx errors (500-511, 530) trigger offline mode
Fixes#1724
This splits indexing into two phases:
* preindex (where only pages, space lua, space styles are indexed)
* index (where everything else is indexed)
The goal is to get to more-or-less operational state quicker this way,
because probably most functionality can already be enabled after the
pre-index phase.
* Align Space Lua arithmetic model with standard Lua
* Add missing __unm and __bnot metamethods
- add new basic tests
- add helper for negative tests
* Add comments for ASCII numeric constants
Signed-off-by: Matouš Jan Fialka <mjf@mjf.cz>
* Don't require auth for .ping
* Finally introducing loggers
* Remove file list caching from disk_space_primitives
I don't think this is required anymore with the sync model
* Simplification of space primitives and more tests
* More reliable online status
* Add .zed settings
* Remove selfUpdate argument from `writeFile` no longer used
- Add `sync` config schema and document sync options
- Move sync config handling to client boot logic
- Move BootConfig type to ui_types.ts
- Remove sync env vars from server
- Update documentation for sync and architecture
* Improve sync logic and proxy routing for full sync state
* Add BoxProxy for late-bound client and improve Lua boot
- Add isMobile syscall for editor
* Add concurrency to sync process and log indexing duration
* Better service worker lifecycle management
* Fix navigate
When two navigations happen in quick succession (e.g. when using the index page navigation hack), one navigation could get stuck
* Fix headers with `from == 0`
Javascript things ahahah
* Fix `#boot` property for PWAs
* [Major] Introduce a new `Ref` and `Path` type
This introduces a new `Ref` and `Path` type, which try to be as lightweight and rigorous as possible. More description in the PR
* [Minor] Updated Syntax
* [Minor] Updated Type
* [Minor] Renamed file
* [Minor] Updated type
* [Minor] Updated Syntax
* [Major] Rename css classes for wiki links
Previously all the parts of a wiki link were tagged with `sb-wiki-link-page`. This didn't make a whole lot of sense. Also the tag `sb-wiki-link` was could never render, because of the way `HighlightStyle` works. Now every part of a wiki link will be tagged with `sb-wiki-link` (i.e. the link, the alias, the dimensions). Invalid and missing links will be styled separately in a later commit.
* [Minor] Adjust to new API
* [Minor] Small changes
* [Minor] Function and file renames
* [Major] Syscall adjusments
Change behavior of `getCurrentPath` to actually return what's now considered a `Path`. `getCurrentPage(Meta)` now also correctly returns for documents. We could consider a rename to get rid of the `page` association
* [Major] Rework `Client`
Mostly minor adjustments for the new API. Some major changes in `navigateWithinPage` and some unecessary checks removed now
* [Major] Errors on navigate
This now throws an error when the user tries to navigate to a link that is invalid
* [Major] Minor adjustments and remove some old bulk
This does the typicall adjustments to the new API. A thing to consider is that invalid links aren't indexed. Also removed old code about template and query stuff. And reused a regex instead of copy pasting it.
* [Minor] Cleanup editor ui
* [Minor] Adjust to new API
* [Minor] Adjust to new API and reduce ugly regex
* [Minor] New API
* [Minor] Rework link cm plugin
* [Major] Rework wiki links
A lot of code cleanup, reworked for new APIs and also introduces "invalid links"
* [Minor] Part of the editor ui rework
* [Major] Adjust to new API
This also removes the `looksLikePathWithExtension` check. I don't think it was necessary.
* [Minor] PArt of editor ui rewrite
* [Minor] Adjust to new API
* [Major] Adjust renderer to new API
Invalid links now link to `#`
* [Major] Rewrite navigator
This still needs major rework, currently in a "got it working" state
* [Minor] Adjust file name
* [Minor] Mistake during rebase
* [Minor] Navigate to end of header
* [Minor] Remove `tweakEditorDOM` and use proper CM extensions
* [Major] Heavily clean up the navigation code
* [Minor] Make `PageNavigator` closable again
* [Minor] Small fixes to reduce errors bubbling up
* [Minor] Show error message if navigate fails
* [Major] Remove legacy `navigate({ kind = ` syntax
This removes the old navigation syntax. It's still supported but the user will be warned then using it. I also added some validation to editor.navigate for user provided refs.
* [Minor] Make notifications errors
* [Minor] Don't fail if header isn't found
* [Minor] Typo
* [Major] Remove cleanPageRef
* [Minor] Small bug
* [Minor] Align behavior between render and editor
* [Major] Remove `PageRef`. Pretty sure it's obsolete
* [Major] Remove pos from header as it's an invalid ref
* [Minor] Align with original behavior
* [Minor] Get tests working again
* [Minor] check that indexPage is valid
* [Major] Rewrite `resolve.ts` file
* [Minor] More tests, more fixes
* [Minor] Don't just append .md this is weird
* [Minor] More edge cases
* [Minor] Version bumping hono somehow fixes the `/` case
* [Minor] Smaller TODOs
* [Major] We can be a littler looser on the restrictions
* [Major] Change header indexing, to index with pos
* [Minor] More fixes for headers
* [Minor] Rename for clarity
* [Minor] Change more regexes to named groups
* [Major] Inline content rewrite
* [Minor] Fix TODO
* [Minor] Bump index version
* [Major] New Space Primitive to check paths
* [Minor] Fix#1214
* [Minor] Fix#1169
* [Minor] Remove TODO
This check is probably not needed anymore. Unsure. May look into it later
* [Minor] small bug
* [Minor] Not a good rebase without me needing to fix it afterwards
* [Minor] Improve fallback for file upload
* [Minor] Some fixes on the for resolving
* [Minor] Removed doubled test case
* [Minor] Resolve#860
* [Minor] Turn flash notification into an error
* [Minor] Change handling of `.` paths
* [Minor] Show empty URL for index page
* [Minor] Push correct ref to history
* [Major] Capture mini editor events in capture phase
This fixes the issue were when pressing arrowup or arrowdown changed the cursor position inside the editor
* [Major] Add prefix to `FilterList`
This now fixes issues were if you had a prefixed page you con't properly search for it, because the search considered the prefix as part of the name
* [Minor] A little rigor for empty headers
* [Major] Fix#1404
* [Minor] `parseInt` returns `NaN` handle that
* [Minor] Use `fileName`
* [Minor] Fix comment
* [Minor] Fixes
* [Major] disallow `foo.png.md` and `/../`, etc. cases
* [Major] Change name of conflicted pages
* [Minor] Remove unecessary check
* [Minor] Remove some split code
* [Minor] Allow deletion of invalid files
* [Minor] Bug in the regex
* [Minor] Give some possibility of fixing incorrect names
* [Minor] Tiny bug with header completion
* [Major] Docs
* [Minor] Ignore errors for navigation inside of a page
This could be a problem when SB e.g. cached a scrollTop for a page, which, because e.g. the file is now deleted, doesn't exist anymore and thus CM throws an error, when trying to navigate
* Update conflicted copy regex in Maintenance.md
---------
Co-authored-by: Zef Hemel <zef@zef.me>
