***ChanServ sets mode: +o rekado_
***rekado_ is now known as rekado
<dannym>janneke: I've put add.c, multiply.c and compile into /tmp on novena (also in my homedir under ~/src/integer-constant/). It seems that this failure (similar to the one I get when building guix wip-arm-bootstrap's glibc-mesboot0-2.2.5) is very similar to something where a lot of stuff was commented out in the preprocessor before <dannym>janneke: Is wip-arm-bootstrap current? <dannym>noadd.c:1: warning: integer constant out of range <dannym>Also, for multiply.c I get a division by zero somewhere <dannym>(to test, invoke "./compile add.c" or similar) <pder>stikonas: do you know if coreutils is needed before musl? I am wondering if we can just build this once after musl is built. <stikonas>and in any case, bash building thing uses coreutils a lot <stikonas>so I think we'll need to build it twice... <stikonas>I briefly tried to play with running some configure scripts, so those 4 are mostly enough, some need date, but we can probably create "fake" wrapper that prints unix time 1 <pder>I have a Makefile now that adds expr, uname, sort, and uniq <stikonas>yeah, my experience was that once I managed to build a few utils, rest were much easier ***dongcarl8 is now known as dongcarl
<fossy> gforce_de1977 please see my most recent comments <janneke>what is the status of powerpc and risc-v in stage0/mescc-tools? <vagrantc>not sure how well it runs, but what could possibly go wrong? <gforce_de1977>janneke: it is on my todo-list, to build minimal qemu-images for all qrchitectures (live-bootstrap as far as possible) <janneke>vagrantc: ah, okay, but it could just cross-build to x86 then ;-) <gforce_de1977>janneke: ofcourse this test will be done, to see if the binaries are identical <stikonas>gforce_de1977: I think fossy wants tiy to move that mkdir out of if to make code simpler. Even if it's only needed in else case, it will be no-op in if case, so harmless <gforce_de1977>fossy: why is 'test condition || bad-case' a 'badstyle'? IMHO it is bad style to do 'test ! condition && bad-case' <cle1408>This video to help people to have a better world without racism or discrimination <pder>fossy: I added a comment that I agree to the licensing <fossy><gforce_de1977> fossy: why is 'test condition || bad-case' a 'badstyle'? IMHO it is bad style to do 'test ! condition && bad-case' <fossy>when you do test condition || { block } <fossy>what that is equivalent to is if [ condition ]; then block; fi <fossy>which is what is almost always what is done in shell scripting, rather than using a block and test <fossy>I dont see any reason to use test instead of if...fi, tbh test just look rly odd <xentrac>hmm, isn't test foo || bar equivalent to if [ ! foo ] ; then bar; fi? <stikonas>I usually use || syntax for something short like command || fail. If it's more text on right side then I usually use if block <xentrac>&& is more often useful, especially if you're not using set -e <fossy>that's what i mean as stikonas said <xentrac>you could write my example as three nested ifs but I think it's more readable if it's flat <xentrac>yeah, || { seems like it's rarely a good idea :) <fossy>stikonas: am i missing something obvious here <fossy>now i get the use of line 53 <fossy>but that's really confusing way <xentrac>I think if ! grep would be slightly clearer than grep ... || { <fossy>it's ok.. i don't mind that too much, but the rest of the code uses if, as do all others in the repo, including for failure <xentrac>otoh test -f ./rootfs.sh || { would be a lot more readable as if [ ! -f ./rootfs.sh ]; IMHO <fossy>hm, gforce_de1977 i don't really get what your commit message means about /dl <fossy>why download to sources/dl instead of sources? i knwon downloads could go anywhere depending on what the user specifies but why make it two layers deep by default <fossy>stikonas: thats what I suggested but my suggestion was not taken up <fossy>although gforce_de1977 your if-like syntax has given me a good idea for how to implement conditionals in kaem <fossy>(Only support them the way you did) <fossy>because that is very easy to implement <xentrac>you can rewrite a ? b : c, when executed for side effects, as a && (b; True) || c <xentrac>when you care about the return value of b, it is not so easy <stikonas>but what are the advantages of all this over sha256sum -c ../../SHA256SUMS.sources ? <stikonas>can't we just run this one command after everything is downloaded? <stikonas>(and maybe add some check in download function that file is mentioned in SHA256SUMS.sources) <fossy>stikonas: yeah, idk.... I am reasonably unconcerned but I dont see any real reason to do it this way <stikonas>fossy: one (minor) con of doing it this way is that we are much stricter on SHA256SUM file structure than even sha256sum is... <stikonas>although, we should still keep style of checksum file consistent... <stikonas>but sha256sum --check does not insist on two spaces <stikonas>ok, I wrote my suggestion there in the comment ***ChanServ sets mode: +o rekado_
***rekado_ is now known as rekado