Repository navigation
Sally 1.0.27: the Unicode and long-path milestone (experimental) #109
0xeb
announced in
Announcements
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Important
Sally 1.0.27 is now available for x64, x86, and Windows on ARM:
https://github.com/0xeb/sally/releases/tag/v1.0.27
Hi everyone,
Sally 1.0.27 is the largest internal update Sally has received so far.
This release completes a major migration of Sally’s core and bundled plugins to native Unicode text and dynamically owned long paths. The goal is simple: Sally should work with the real file name you selected, regardless of the Windows system locale or active code page.
Warning
Sally 1.0.27 should be considered experimental.
This change touches file operations, panels, dialogs, search, viewers, editors, archives, plugins, configuration, shell integration, and many other internal boundaries. It has received extensive automated and manual testing, but regressions and unexpected compatibility problems are still possible.
Please keep Sally 1.0.26 available until you have tested your normal workflows. If Sally is critical to your work, do not immediately replace your only trusted installation.
What has changed
MAX_PATH-style internal limits.Behavior that may look different
Sorting now compares actual Unicode names instead of damaged code-page projections. This means files with non-ASCII names may appear in a different order than in older versions. That is intentional, although further sorting feedback is welcome.
Some very old third-party programs or external tools that only accept ANSI paths may reject names they cannot represent. Sally should now report that failure instead of silently changing the path and potentially targeting a different file.
Older third-party Sally plugins remain supported through an isolated compatibility layer, but plugin-specific regressions are possible after a migration this large.
I need your help testing it
Please test the workflows that matter to you, especially:
If something worked in 1.0.26 and fails in 1.0.27, please open an issue and include:
Help > Report a Bugoutput when available.This migration has already uncovered and fixed many old assumptions that only appeared outside English/Western file names. Real-world reports are the best way to find the remaining ones.
Thank you to everyone who has tested Sally, reported issues, and helped make this milestone possible. The experimental 1.0.27 release is available for x64, x86, and Windows on ARM from the release page linked above.
All reactions