Bug fixes in dnd.c:
- mktempdir_in_cache: add missing 'return ans' after successful strdup
- mktempdir_in_cache (utils.py): use O_RDONLY instead of O_RDWR for directories
- remote_items allocation: use mi.num_uris instead of ds.num_mimes
- Off-by-one: uri_item_idx > changed to >= for bounds checks
- Off-by-one: entry_num > changed to >= for bounds checks
- populate_dir_entries: fix missing last entry after final null separator
- add_payload directory finalization: create directory on disk with mkdirat
- get_errno_name: add EFBIG, EISDIR, ENOSPC error codes
Test infrastructure:
- Add dnd_test_force_drag_dropped() to simulate DROPPED state
- Make notify_drag_data_ready() succeed in test mode
Comprehensive t=k tests added:
- Single file, empty file, single symlink transfer
- Chunked file transfer with m=1
- Single directory with children
- Directory with symlinks
- Multiple URIs
- Deep directory trees (breadth-first and depth-first, 3+ levels)
- Mixed file/dir/symlink at top level
- Completion signal
- Error handling (client errors, invalid state)
- DoS limits (REMOTE_DRAG_LIMIT, PRESENT_DATA_CAP)
- Invalid input (bad base64, too large chunks, invalid indices/handles)
- URI list with comments
- Multiple chunks for directory listing
Agent-Logs-Url: https://github.com/kovidgoyal/kitty/sessions/9da0bff7-6a1a-490f-a4c5-8cb328e056ce
Co-authored-by: kovidgoyal <1308621+kovidgoyal@users.noreply.github.com>
This was needed to fix various corner cases when doing blending of colors
in linear space. The new architecture has the same performance as the
old in the common case of opaque rendering with no UI layers or images.
In the case of only positive z-index images there is a performance
decrease as the OS Window is now rendered to a offscreen texture and
then blitted to screen. However, in the future when we move to Vulkan or
I can figure out how to get Wayland to accept buffers with colors in
linear space, this performance penalty can be removed. The performance
penalty was not significant on my system but this is highly GPU
dependent. Modern GPUs are supposedly optimised for rendering to
offscreen buffers, so we will see. The awrit project might be a good
test case.
Now either we have 1-shot rendering for the case of opaque with only ext
or all the various pieces are rendered in successive draw calls into an
offscreen buffer that is blitted to the output buffer after all drawing
is done.
Fixes#8869
No longer pay the overhead of starting the Python interpreter
just to write a message to the single instance socket. This reduces
the time for kitty --single-instance (for second and later instances)
from 70ms to 3ms an almost 25x improvement.
Needs testing on macOS and also porting of the +open handling.
Follow PEP 0394 and use /usr/bin/env python so that the python in the
users venv is respected. Not that the kitty python files are meant to be
executed standalone anyway, but, whatever.
Fixes#6810
For instance the user could have
[ "$KITTY_HOLD" = "1" ] && exec kitten __hold_till_enter__
in their shell rc files to get back the previous Press enter or esc to
quit behavior.
