IRC channel logs
2023-02-13.log
back to list of logs
<stikonas[m]>hmm, building GCC 10 in live-bootstrap really takes quite a while... <stikonas[m]>fossy: I wonder if we can try enabling parallel build for some packages <stikonas[m]>guile and gcc and python might particularly benefit from it <stikonas[m]>and GCC 12 will be even worse if we do --enable-bootstrap for rebuild with itself <stikonas>I get errors like: "/usr/bin/ld: /usr/lib/i386-unknown-linux-musl/gcc/i386-unknown-linux-musl/10.4.0/libgcc.a: error adding symbols: archive has no index; run ranlib to add one" when trying to build Hello World <fossy>stikonas[m]: i was also thinking abt parallel build, we do need to visit that soon <fossy>i didn't test the very very last iteration of GCC. but it got good testing before that - but most of that was in-tree <fossy>im sure i tested simple things out of tree... <fossy>.. cause in tree, nothing is stripped, and i didn't test out of tree since i added stripping right at the end <fossy>might just kill the stripping for now, and apply it globally in a little bit <fossy>before gcc 12, gonna take a detour to impl parallelism and global binary stripping <stikonas[m]>fossy: gentoo. actually uses scanelf from pax-utils for detection <Piraty>how big of a donation one must make to the project to be assured nodejs will not be included in bootstrappable chain and thus not be part of $post_nuclear_world ? <oriansj>Piraty: one needs only not do that work and hope no one else has an idea to address it <Piraty>that thinking is dangerous, as it didn't save us from it in the first place <oriansj>Piraty: you seem to be under the mistaken belief that there will only be one true bootstrap chain. The truth is a bit more complex. Who does, decides. Means there are no leaders to bribe, saints to consult or developers to blackmail. No one get the say of what does or does not get developed or how it gets developed, except the very people doing the hard work. You by all means can write checks to each and every single developer here in <oriansj>exchange for them promising not to do that work but any software developer with a strong desire could build that link. So you ultimately could have written all those checks for nothing. It also means, I don't have to import other people's work and they don't have to import mine either. <doras>fossy: I think you forgot to mention file-5.44 in parts.rst <oriansj>so Piraty if you want a bootstrap chain with none of the node.js work; you need only do the work required (which right now is literally zero work required) <oriansj>So just working on live-bootstrap is a good start; and if you want to buy even comfort in that live-bootstrap will not add the node.js work. I'm sure fossy and stikonas are more than happy to recieve money on the basis of not including node.js into live-bootstrap (and someone else will probably just create live-bootstrap-node.js that just builds upon live-bootstrap) <stikonas>but anybody can bootstrap distro off live-bootstrap <stikonas>and most distros definitely package nodejs <stikonas>but nodejs doesn't seem to be needed for lower level parts, such as getting to GCC <stikonas>fossy used file to strip size of gcc 10 but that ended up with broken gcc <stikonas>oriansj, Piraty: and javascript is the language that I don't really know, so I wouldn't really use it myself <stikonas>even without knowing the language one could see lack of logic... <stikonas>well, I'm not opposed to having some stuff in live-bootstrap that is not strictly necessary for bootsrapping <doras>It's not really a "step" in the same sense, I guess. <stikonas>we usually list those in parts.rst anyway <stikonas>i.e. stuff like makeinfo is completely optional