* fixed problem on S3: turns out to be mem alignment
* fixed scaling for C3
* code cleanup, added arduinoFFT back in, added ifdefs & description
also added major peak and frequency bin calculation for DSP FFT
* moved ifdefs to correct place, separated sample filter & FFT filter application
post FFT band pass and IIR applied to samples are now separated: I found in testing that applying the sample filter helps with aliasing into base-bands, there is no need to hard-cut the lowest frequencies after FFT.
* changed sample low pass cutoff from 80Hz to 90Hz
better anti-aliasing at minimal loss of base frequency.
* code cleanup and minor speed improvement
- moved scaling of FFT values into fftAddAvg() to use ferwer operations
- added "using" for math types, removing some ifdefs and duplications
* new font format: *.wbf with a 12byte header, bit packed data and support for variable char width
* add support to load custom fonts from file system
* UTF-8 to unicode support functions
* support for any unicode offset in char 128-255 enables many international chars
* update built-in fonts with similar but nicer ones
* update pixelforge scrolling text tool with a preview and support for custom fonts
* accompanied by Font Factory tool (pixelforge) to easily create custom fonts from various formats
* removed fastled dependencies
- copied relevant functions
- optimized some of them for ESP32
* added perlin functions from PR, code cleanup. work in progress
* added hsv2rgb16rainbow function, some cleanup
- new rainbow function is faster and more accurate than original fastled function
* updated conversion functions (now faster), cleanup, optimizations
* code cleanup, moved (most) fastled functions into fastled_fcn.cpp
- resolves licensing issue
- moved PRNG into its own space
- prettified some code
* fixed prng: it now generates a full sequence of random numbers, thx @TripleWhy for pointing out the flaw
* rename to fastled_slim, add hsv2rgb() convenience aliases, fix FX usage
* improve ease8inOutCubic() accuracy
* fix background in twinklefox, minor optimization in PS (always use gamma LUT, no ifs)
* bugfixes in FX, adding white and cct transition to slow transition FX
Co-authored-by: Frank <91616163+softhack007@users.noreply.github.com>
Clean-up and simplification of strip.service() after individual effect FRAMETIME was removed.
* Activation "_frametime" is now the same for all effects, so effects scheduling based on individual frametime is not needed any more
* Special handling for mode_static is obsolete, because the solid effect does not have a different timing any more
* Ensures a safe fallback segment after servicing to avoid stale state.
* Brightness changes use a capped refresh threshold to avoid excessive redraws.
* load pixel forge tools from json file, bugfixes
- load available external tools from a json file instead of hard-coding them in
- add buttons to update or delete tools (instead of manually deleting files in /edit)
- fixed bug that file system memory was not updated after uploading or deleting files
- fixed a bug where gifs with a space would not display
- fixed bug of page not loading properly (out of sync access to tabSw() )
- remove pixelforge from ESP8266 1m builds (does not fit)
- add inline html page if pixelforge is not built in to avoid 404
@coderabbitai caught the problem, but its fix suggestion was out to
lunch on this one. Passing an unsigned char to String.concat() appends
it as a number instead of appending the character.
Instead, mask the unneceeded bits.
Discard IPv6 RA packets, which cause LwIP to overwrite the IPv4 DNS
servers, resulting in DNS queries hanging up because there are no
accessible servers.
In startTransition(), a copy of the old effect's data buffer is made using raw malloc() — it deliberately bypasses allocateData(). This buffer is never added to the _usedSegmentData static counter (only allocateData() does that). If the old effect then calls allocateData() it causes _usedSegmentData to underflow.
https://github.com/wled/WLED/issues/5427#issuecomment-4103228702
Bug Fixes
* Improved hostname sanitization and buffer-size safeguards to prevent truncation and ensure DNS-compliant names.
* Hostname is now applied earlier during network setup and on reconnects (ESP32), improving connectivity and ensuring settings take effect.
* Solves the problem that ESP32 boards show as "ESP-xxxxxx" on the network, no matter if DNS name is defined or not
* (breaking fix) prevent hostnames with double ``WLED-`` prefix, like ``wled-WLED-kitchen``
New Features
* Added a length-aware hostname retrieval API with an mDNS-preference option and robust fallback behavior.
* Preserved legacy compatibility by providing a length-aware alias for legacy hostname preparation.
Open Ends
* mDNS name is not sanitized when registering the mDNS endpoint, to avoid breaking previous functionality
Make AsyncDNS queries return a future-like object backed by a smart
pointer. This ensures that the client cannot release the backing
memory while the dns query callback is still holding a reference to it,
preventing any use-after-free cases.
* use same style as other settings pages
* Hide "back" button while update is in progress
* set "download latest binary" URL and badge based on info.repo; directly link to "latest" release
* correct bad name of "Security & Updates" page
* ensure that "update bootloader" section get hidden when not supported
Updated instructions regarding generated html_*.h files to clarify that they should not be committed, and they don't need to be re-generated in any PR.
* Info page updates and minor re-styling
* added GitHub repo (link)
* added Total LEDs
* removed lwip major version on esp32
* two horizontal lines for better readability
* add rel="noopener noreferrer" for improved security
* When using target="_blank", it's a security best practice to include rel="noopener noreferrer" to prevent the new page from accessing window.opener.
