IRC channel logs
2023-07-15.log
back to list of logs
<pder>doras: interesting. I tried downloading gawk-5.2.2 source and building with gcc 13.1.0 and it does not segfault <pder>I wonder if --enable-default-pie is causing any issues <doras>pder: we're building gawk-5.2.1, but it likely doesn't matter. We're also building it with `--prefix=/cross` and `--build=i686-bootstrapper-linux-gnu` and `--host=i686-bootstrapper-linux-gnu`, if it makes a difference. <doras>And CFLAGS=-g0 CXXFLAGS=-g0 LDFLAGS=-Wl,-z,relro,-z,now -Wl,--as-needed exported in the environment. <stikonas>mihi: yes, we want to use mes for gash too but probably not much more <oriansj>stikonas[m]: I think we are entering the region of investment and return; which may indicate a previously unexamined assumption about scheme and long term support costs <fossy>doras: thanks for that segfault information, i'll look into that <fossy>gawk seems to be simplest example <fossy>mihi: i do not know for certain, actually. that is all i could find - but i'm no scheme expert, so there could be more.. <fossy>stikonas: i don't think it's impossible to rewrite it in C, but it is very scheme centric right now, as in the entire parsing etc results in a scheme object, so mes would probably also need a bit of rewriting <doras>Was Linux detected correctly only in one of these cases? Do you know why? <doras>I guess it suggests that we may have some gap between the two that could be worth investigating. <fossy>doras: i do remember that, it was that the default MACHDEP=linux$LINUX_VERSION_MAJOR so eg, my chroot had MACHDEP=linux6 whereas QEMU had MACHDEP=linux4 <doras>fossy: oh, so it's a matter of kernel version. <doras>Yeah, not a lot we can do about it... <fossy>kernel version is one of the few things where we can't really get around it.. <stikonas[m]>indeed, even if we can patch our bootstrap kernel in live-bootstrap, we can't patch kernels that everybody else uses\