13 Oct 2025 |
Marie | anything else i could show to debug this? | 19:11:05 |
raitobezarius | the FOD result | 19:11:28 |
raitobezarius | if you can construct it somehow in another way, great | 19:11:36 |
Marie | i did | 19:11:41 |
raitobezarius | i'm so dumb | 19:11:47 |
raitobezarius | it'swrong | 19:11:55 |
raitobezarius | ok cool | 19:12:02 |
raitobezarius | it should be an easy fix | 19:12:05 |
raitobezarius | something very going is happening with regexes | 19:12:10 |
raitobezarius | std::string LinuxLocalDerivationGoal::rewriteResolvConf(std::string fromHost)
{
if (!runPasta) {
return fromHost;
}
static constexpr auto flags = std::regex::ECMAScript | std::regex::multiline;
static auto lineRegex = regex::parse("^nameserver\\s.*$", flags);
static auto v4Regex = regex::parse("^nameserver\\s+\\d{1,3}\\.", flags);
static auto v6Regex = regex::parse("^nameserver.*:", flags);
std::string nsInSandbox = "\n";
if (std::regex_search(fromHost, v4Regex)) {
nsInSandbox += fmt("nameserver %s\n", PASTA_HOST_IPV4);
}
if (std::regex_search(fromHost, v6Regex)) {
nsInSandbox += fmt("nameserver %s\n", PASTA_HOST_IPV6);
}
return std::regex_replace(fromHost, lineRegex, "") + nsInSandbox;
}
| 19:12:22 |
raitobezarius | this is how the rewrites should take place | 19:12:26 |
raitobezarius | if anyfew has some time to stare at this and find the bug, it'd be greatly appreciated | 19:12:45 |
raitobezarius | (as I'm wrangling with profiling stuff rn) | 19:12:49 |
Marie | I'll have a look | 19:15:03 |
hexa | uhh, I don't know pasta, but how does it map from PASTA_HOST_IPV4 to the host resolver or wherever? | 19:28:57 |
hexa | especially if the host resolver points to loopback | 19:29:09 |
hexa | this only substitutes nameservers from the host resolv.conf from what I see | 19:35:43 |
hexa | so as long as it maps the PASTA_HOSTA_IPXX addresses it should be fine? | 19:36:04 |
raitobezarius | pasta connects things together | 19:37:04 |
KFears (burnt out) | Same thought... Resolution should be independent from connectivity | 19:37:05 |
raitobezarius | so the PASTA_HOST_IPV4 is exactly 127.0.0.1 in the other world | 19:37:12 |
raitobezarius | pasta host IPs are AIPA or a ULA that we generate IIRC | 19:37:35 |
raitobezarius | * pasta host IPs are AIPA (v4) or a ULA (v6) that we generate IIRC | 19:37:41 |
K900 | Can it not just do both | 19:38:18 |
K900 | And let the applications happy eyeball | 19:38:25 |
raitobezarius | it perhaps should | 19:38:49 |
Yureka (she/her) | Is it correct that deleting store paths from a http(s) binary cache is pretty much impossible without destroying the user experience? | 19:39:15 |
raitobezarius | what does that mean? | 19:39:26 |
raitobezarius | you mean that deleting a random path in the middle lead to explosions? | 19:39:43 |
Yureka (she/her) | No, deleting a whole isolated closure | 19:39:50 |