IRC channel logs

<stikonas>fossy: just noticed your perl PR

<stikonas>quite a lot of intermediate steps...

<stikonas>but at least nothing really early, all the perl 5.6 stuff was alright

<fossy>stikonas: yeah, it is very unfortunate

<fossy>i did try to make it as efficient as possible, and it probably _could_ be made more efficient, but that would require a lot of effort

<fossy>yeah, i am also glad early perl is okay, that would have made life a lot worse

<fossy>Thanks TCC for very sensible behaviour

<fossy>they define a symbol "unix" to be NULL on a unix system

<fossy>so you can't have a variable named unix

<fossy>:D

<fossy>that makes no sense

<lanodan>Huh that probably makes it a non-conforming compiler as it's effectively namespace pollution

<fossy>yeah....

<matrix_bridge><cosinusoidally> doesn't gcc also define "unix" (at least when it's in gnu89 or gnu99 mode)?

<lanodan>Well in ISO C modes at least it should be __unix or __unix__

<lanodan>Or C89 aka ANSI C.

<stikonas>fossy: strange: there is no strings "unix" in tcc source code

<stikonas>only __unix__ and __unix

<Googulator>vasi: got the ccache-related error again, in package "libkcapi"

<Googulator>...ok, looks like ccache is a red herring

<Googulator>the actual error, seen in the logs, is a test failure: "Allocation of hmac(sha512) cipher failed (ret=-97)"

<Googulator>which makes little sense - errno 97 is "Address family not supported by protocol"

<Googulator>...unless it's something related to the kernel configuration - libkcapi is a userspace library for accessing the kernel cryptography APIs, after all

<Googulator>turns out, the entire kernel crypto API export to userspace was disabled in kconfig

<Googulator>kconfig updated - vasi: you will need to rebuild the 64-bit kernel (assuming you aren't starting from a clean live-bootstrap)

IRC channel logs

2025-09-10.log