| 16 May 2026 |
 Randy Eckenrode | End of Wine updates is just the nature of Apple’s platform transitions. | 18:40:55 |
| Randy Eckenrode | The problem with running Wine natively is explained in that Bugzilla issue I linked on Discourse. macOS does not (normally) all ARM64 processes to use the X18 register or map memory in the lower 1 GiB. Both are needed to support Windows ARM64 applications. | 18:41:06 |
| Randy Eckenrode | Apple hasn’t signal whether or how they will try to accommodate that. If they do, my guess is it will require a special entitlement. | 18:41:58 |
| Randy Eckenrode | We also don’t know how Rosetta 2 will support only ‘older`, ‘unmaintained’ games. For all we know, they’ll run a version of macOS 27 in a VM. | 18:43:12 |
| Randy Eckenrode | At some point, I expect Apple to stop supporting TSO in hardware for Intel emulation. Rosetta 2 will eventually be dropped. I would be surprised if Apple maintains it indefinitely. | 18:44:40 |
 emily | right, but if you don't care about ARM64 applications then that's not necessarily an obstacle to running x86 ones, right? | 18:49:36 |
| emily | I imagine CodeWeavers cares very little about running actual ARM64 Windows apps, because there basically aren't any | 18:49:55 |
| Randy Eckenrode | What can an attacker control? The most common uses are playing games and running Windows-only applications. An attacker would have to convince a user to run untrusted input that exploits both Wine and its native libraries, or you might attack the user’s computer if you can get their IP address. If you’re socially engineering the user, you might as well just install regular old malware. | 18:50:31 |
| emily | I do wonder if they're going to make any judgements to GPTK | 18:50:39 |
| Randy Eckenrode | (Maybe users might have a problem if they’re browsing the open web using a browser under Wine, but that’s nuts.) | 18:51:14 |
| Randy Eckenrode | If DXMT gets good enough DX12 support (or KosmicKrisp can run whatever the Vulkan DX12 layer is), I can see Apple’s retiring the D3D part of GPTK. The one that actually matters is the shader converter, which I assume already supports aarch64-darwin. | 18:53:57 |
| Randy Eckenrode | * (Maybe users might have a problem if they’re browsing the open web using a browser under Wine, but that’s nuts. Why would they do that?) | 18:54:49 |
| Randy Eckenrode | * (Maybe users might have a problem if they’re browsing the open web using a browser under Wine, but why would they do that?) | 18:55:07 |
| 17 May 2026 |
 hexa | can anyone say with certainty what happens to ssh keys after running startosinstall? | 15:08:20 |
| hexa | Redacted or Malformed Event | 15:08:25 |
| hexa | Redacted or Malformed Event | 15:11:02 |
| hexa | Redacted or Malformed Event | 15:11:08 |
| hexa | my assumption would be that we'd be fine if we didn't pass --eraseinstall | 15:15:40 |
| hexa | but I'm not keen on trying it out and being wrong about it | 15:15:50 |
| hexa | Redacted or Malformed Event | 15:16:03 |
| hexa | error:
… while evaluating an expression to select 'drvPath' on it
at «internal»:1:552:
… while evaluating strict
at «internal»:1:552:
(stack trace truncated; use '--show-trace' to show the full trace)
error: infinite recursion encountered
at «internal»:1:490:
note: trace involved the following derivations:
derivation 'zstd-1.5.7'
derivation 'bash-5.3p9'
derivation 'clang-wrapper-21.1.8'
derivation 'cctools-binutils-darwin-wrapper-1010.6'
derivation 'cctools-binutils-darwin-1010.6'
derivation 'llvm-21.1.8'
derivation 'libxml2-2.15.2'
derivation 'libiconv-113'
derivation 'python3-3.14.5'
derivation 'cmake-minimal-4.1.2'
derivation 'darwin-binary-paths.patch'
derivation 'system_cmds-1039'
derivation 'system_cmds-deps-private-headers'
derivation 'AvailabilityVersions-155'
| 21:15:14 |
| hexa | python314 depends on zstd, which is what I assume results in this new infinite recursion on python-updates | 21:15:37 |
| hexa | looking for great ideas how to resolve that | 21:15:55 |
| Randy Eckenrode | Is it a requirement? Can support be disabled in the Darwin bootstrap? | 21:29:43 |
| Randy Eckenrode | Alternatively, we could build zstd in the bootstrap without libiconv support. That might work. | 21:31:32 |
| hexa | yeah, it's kinda a requirement | 21:32:40 |
| hexa | it's one of the stdlib modules now and python packages will break if we disable it | 21:33:00 |
| hexa | and we disable it already on minimal | 21:34:09 |
| Randy Eckenrode | Darwin bootstrap doesn’t use minimal Python because some things expect a more fully featured Python. | 21:34:48 |
| Randy Eckenrode | If we can disable only zstd during the bootstrap, that should be fine. | 21:35:54 |
