IRC channel logs

2021-04-24.log

back to list of logs

<OriansJ>Hagfish: # would be a comment line but the cat would dump to stdout and the || would ensure in cases where builtins/pipsize.h didn't exist that it wouldn't abort the rest of the build.
<OriansJ>bauen1: why do a OS as library when at that stage you have M2-Planet and mescc-tools running on bare metal?
<OriansJ>wouldn't one want to get as much functionality in a straight shot to Linux?
<OriansJ>aka build a POSIX capable of running the steps leading to the building of GCC+Linux?
<OriansJ>I mean, why complicate the bootstrap with having to setup another build environment when you can go straight to getting everything you want done?
<OriansJ>now if one were on a desert island and had to use only the things they bootstrapped, that would be reasonable to a degree. But since have have much more tools to work with, why not take advantage and build the best bootstrappable pieces possible to reduce the total number of steps?
<fossy>is there anyone else who feels like when they step away from doing something and then come back after a few days everything just works
<fossy>because i paused my perl work cause of my exam for school and came back today and found the problem after just half an hour of debugging
<fossy>:D
<fossy>perl 5.32.1 is done!
<fossy>well as long as this compile works, i am 90% sure that it will
<fossy>stikonas: i think i might investigate doing some parallelism... i think our compiles will become very slow very soon if we don't use multithreading where possible
<stikonas>fossy: also I think qemu is really slowing CI...
<xentrac>bauen1: you might not even need a microkernel; a CP/M-style or MS-DOS-style BIOS might be enough
<stikonas>my runs in chroot are much quicker
<stikonas>nice work on perl!
<OriansJ>fossy: it is called the benefits of a fresh perspective
<xentrac>congratulations, fossy! perl is a beast
<stikonas>yeah, perl is a bit annoying due to metaconfig...
<stikonas>not that autotools is much better...
<stikonas>but at least autotools is so common, that there is more knowledge about it
<gef>@fossy: a night's sleep is well known to produce miracles! On one occasion, I even solved one very hard problem *during* sleeping. Alas, it happened only once, it would have been big cheating otherwise ;-)
<fossy>stikonas[m]: yes, i agree that qemu is slowing CI, i mean in general though (perl unparallised takes 5+mins on my machine)
<fossy>i'm considering that we should use chroot in CI
<gforce_de1977>fossy: congrats for perl! (solving problems after having distance is common, solving problems while sleeping is also not unknown ..., today i'am doing a hike, lets see which problems will solve this eventually 8-))))
<gforce_de1977>i had 2 failed CI-builds tonight during massrun: http://intercity-vpn.de/bootstrap/bootstrap.log-multilog-19-1619235591.txt
<gforce_de1977>sad: my auto-base64-dumper did not work, have to find the mistake
<gforce_de1977>on the otherside: using 'find' after a failed 'findutils' build will maybe not lead to success 8-)))
<stikonas>fossy: I guess let's switch to chroot then...
<stikonas>(for CI)
<stikonas>it will also gives us more flexibility for which CI runner we use
<OriansJ>gforce_de1977: xeh exists in stage0-posix for a reason, so you can always look at the hex of a file
<gforce_de1977>OriansJ: wow! xeh is new to me, very good
<gforce_de1977>(unsure if we build that, will check)
<gforce_de1977> https://github.com/oriansj/stage0-posix/blob/master/Legacy_pieces/xeh.hex0
<OriansJ>gforce_de1977:it along with sin can be used to audit any source code file you have which is trying to play unicode games