This is a meta-issue to track progress on bootstrapping pkgsrc on SerenityOS. While I don't expect it to replace the native Ports systems, it's a good stress test of POSIX conformance, performance and stability.

Blocked on:

• #8473 Storage stack issues with large partition sizes
• #8567 Various small POSIX compliance issues
• #8506 expr POSIX compliance, needs non-trivial tweaks to LibRegex
• #8613 expr POSIX compliance, bug fixing
• #8617 expr POSIX compliance, doesn't return the correct value for a matched subexpression
• #8704 Missing re_nsub field inside regex_t structure
• #9330 Breaks bmake
• #9343 Missing sigsuspend implementation for mksh
• #9344 Missing various tr functionalities for pkgsrc's bootstrap scripts
• #9345 C library doesn't contain the POSIX C regex API (regerror, regexec, regcomp...), bad default PATH (is /bin:/usr/bin:/usr/local/bin, should be /usr/local/bin:/usr/bin:/bin), missing cc symlink when installing gcc port, missing awk symlink when installing mawk port
• #9257 Running out of space during image generation with pkgsrc
• #9392 Erratic mksh behavior (won't quit, burns the CPU on EOF)
• #9442 ls -t is broken, cp -p isn't implemented
• #9604 Yet more sigsetjmp fails
• #10077 #10097 sed port is broken because wcscoll() and others are TODO() placeholders
• #10594 ln, touch and FOPEN_MAX fixes
• #9268 The port itself
• ...

Tasks:

• Document how to bootstrap pkgsrc on SerenityOS (done in #9268)

Bonus:

• grep is missing the -c/--count option (can be worked around with grep from the ports)
• Shell isn't POSIX-compliant enough to run bootstrap script (can be worked around with bash from the ports)
• Investigate performance issues with large directories (>3000 entries) because of AK::String allocation inside Kernel::Custody (uses kmalloc() internally, burns CPU)
• Investigate performance issues with large amount of process spawning during configure script execution
• Investigate why CPU usage rarely seems to exceed ~35% while running configure scripts
• Investigate stability issues (occasional random crashes related to ProcFS), especially during fork-heavy loads alongside things accessing ProcFS
• Investigate some kind of issue surrounding waitpid() & co. (sometimes bash processes get stuck in the Waiting state without any child processes to wait on, can be worked around most of the time by killing the stuck process)
• Investigate what the heck happens when a configure script tries to find out the maximal length of command-line arguments (deadlocking bash process while writing a diagnostic message during an assertion triggered inside free()?)