IRC channel logs
2022-04-08.log
back to list of logs
<oriansj>unmatched-paren: is that really a surprise? bootstrapping is a marathon and not a race <oriansj>you can always pick up where someone else left off and drive it into new and exciting directions. <oriansj>with a good history of releases and no malicious jumps (aka needing the functionaly of the release to build the release sort of crap); everything can be bootstrapped by just doing some history but at this point we now *KNOW* that is rarely the case <oriansj>so I guess one could say it is mostly off-roading for bootstrapping. <oriansj>The roads have all decayed to rubble and we have to fix them up and keep them in working order. <muurkha>fortunately it only rains once every 20 years <oriansj>muurkha: overly optimistic today I see <stikonas>fossy: I'm getting a bit stuck with my /after->/sysa rename... <stikonas>not sure why but perl-5.32 is not deterministic <stikonas>or probably just different for qemu vs chroot builds <Hagfish>it would be annoying if it is non-deterministic, but at least the build environment hypothesis is a testable one <stikonas>well, first of all I need to compare tarballs... <stikonas>at least with qemu having disk image it's now easier to extract tarballs from qemu build <markjenkinssksp>My public mainline doesn't show it, but I was actually active until end of December in a private branch which I just squash merged and made public. Got into big trouble where my gcc builds were showing serious signs of memory errors bugs, but my M2-Planet builds were not. <markjenkinssksp>Frustration with that made the project easier to shelve in the face of competing personal and professional demands. But I always kind of knew in the back of my mind that Valgrind could probably show me where I went wrong. And yesterday it sure did, hence the above commit where I figured out my attempt to do pointer arithmatic in a manner that works for both gcc and M2-Planet was never correct for gcc. <stikonas>markjenkinssksp: by the way, these days you can use slightly more features and still be compatible with M2-Planet <oriansj>markjenkinssksp: also we now support #ifdef blocks in M2-Planet <oriansj>so you can do #ifdef __M2__ CELL_SIZE = sizeof(Value) #else CELL_SIZE = 1 #endif <stikonas>Hagfish: wow, the diff between chroot and qemu is really strange <stikonas>so pod .file in chroot mode is identical to the one in source tarball <oriansj>the line number is rather interesting for such a change to occur <markjenkinssksp>cool to know there are new features. Will be interesting to see if my source as is has remained M2-Planet compatible <Hagfish>lol, i've just seen the diff between chroot and qemu outputs. yeah, that is really cursed <littlebobeep>I think this might be the coolest chat room I have ever found <oriansj>markjenkinssksp: just a minor warning about *var = blah; it isn't fully ISO standard but it does work on the trivial case as it is compatible. <Hagfish>littlebobeep: it's even cooler now you're here :)