this is going to be my response whenever anyone questions the concept of directory hardlinks in the future https://tldr.nettime.org/@tomjennings/117390970337803411 it is a necessary primitive for abstraction of filesystem contents for the specific purpose of translation across a serialization boundary
0 Boosts
0 Likes
then i found it was impossible to be correct because of the intentionally fucked up POSIX VFS API which COMMITS THE UNFORGIVEABLE INODE BETRAYAL and resolved to make my own filesystem. then i realized transactions make everything easier not harder and my hot anger turned to cold hate in an instant of reverse sublimation
then i fixed a tiny documentation typo in findutils, and they were very nice to me https://lists.gnu.org/archive/html/bug-findutils/2025-08/msg00001.html you will find this is why i like gnu projects it's because they're actively nice to me
so i have studied every form of the blade across history. i am no longer a disciple of the blade. i am beyond mastery. i am the blade
i spent a year and a half studying parallel recursive directory traversal, made a literally perfect and #safe path handling library with three novel concepts worth publishing in their own right, then constructed an abstract dataflow graph that enables literally arbitrary user logic at any point of the traversal, which nobody else even understands enough to care about yet. this resulted from finding a deadlock in ripgrep and then digging and digging and digging https://github.com/BurntSushi/ripgrep/pull/3010
ignore: gracefully quit worker threads upon panic in ParallelVisitor by cosmicexplorer · Pull Request #3010 · BurntSushi/ripgrep