<korvo>
I'm told that this general path, with the "minimal" PyPy, works fine on actual aarch64 hardware. That makes me unworried about any actual users of rpypkgs, but it does eliminate all of my ideas about why CI would fail.
<LarstiQ>
korvo: know as in I do run it, but not what that error might mean
<korvo>
LarstiQ: No worries, thanks.
nightstrike has joined #pypy
infernix has quit [Ping timeout: 252 seconds]
infernix has joined #pypy
<tumbleweed>
korvo: in Debian we build pypy 3.x with cpython 2.7, you could try that approach
<tumbleweed>
(avoiding that bug)
itamarst has joined #pypy
nightstrike has quit [Quit: Connection closed for inactivity]
jcea has joined #pypy
<korvo>
tumbleweed: Hm. I had thought that I was already doing that; this bootstrap path starts from its own CPython 2.7 which isn't shared with the rest of the system. Is there something aarch64-specific going on, or is it a known version issue?
<korvo>
I appreciate the advice though. I wouldn't trust Ubuntu's versions either.
<tumbleweed>
korvo: the failure you were getting was in translating pypy2.7
<korvo>
tumbleweed: Oh, I see what you mean. Yes. But if this one fails then all of the others are probably going to fail as well... The flake builds PyPy 2.7 with `--translationmodules --withmod-thread`, which should build a PyPy that can't run most user-level stuff but can bootstrap PyPy.
<korvo>
Otherwise it's `--allworkingmodules`, which I think enables a lot more stuff.
<korvo>
tumbleweed: Okay, let's see what happens. I've patched the bootstrap to just not build minimal PyPy 2.7, and to use the bootstrapping CPython 2.7 for everything. A fresh check with bootstrap takes ~1h; I don't know how long this will take but I'd guess ~3h.