***ng0_ is now known as ng0
***OriansJ` is now known as OriansJ
<OriansJ>I don't know about anyone else but this week is just kinda making me feel completely drained of energy. Not to mention the whole rms/fsf thing <Hagfish>yeah, several people have said this week has been a write-off for them <janneke>this has been a very unusual week for me too <Hagfish>i guess the aliens in Area 51 really don't want us to find them ;) <OriansJ>Hagfish: well Area 51 certainly has a kick-ass lazer-tag range; I find their pizza rather embarassingly bad <OriansJ>janneke: my thought is to do blood-elf in scheme this weekend and M1 next week <janneke>OriansJ: that's great. i'm making some slow but nice progress with gash' core-utils, mes and the path beyond <civodul>janneke: BTW, since we're finally close to merging core-updates, what would you think of writing a blog post to explain this? <civodul>i know it's not exactly fresh news for you, but hey ;-) <OriansJ>civodul: do you think it will be a good time to introduce stage0 as part of the future of the guix bootstrap seed? <OriansJ>dddddd: any news on how the aarch64 port rewrite is going? <janneke>OriansJ: we can always start to prototype it. <janneke>i always thought introduring stage0 would take much longer, but it seems we're getting really close? <janneke>as civodul mentions, we're at the point of merging the reduced binary seed bootstrap, that brings the bootstrap binary size from 250MB down to ~130 <janneke>the new scheme-only bootstrap that i hope to have ready in half a year or so (and possibly merge a year from now) should reduce that to ~60MB (basically: guile, mes and mescc-tools) <civodul>i feel bad that it's taken us sooo long to merge it <civodul>it's clear that our workflow for handling core-updates leaves... ahem, room for improvement <OriansJ>janneke: well hopefully the scheme-only bootstrap will reduce down to just the guile binary, once I complete slow-utils <janneke>i haven't dared to look at how well gash runs on mes yet, that will probably need some work <OriansJ>janneke: well the gap right now is mes-m2; which when completed solves the stage0 to MesCC bootstrap (assuming the self-host is able to do the full bootstrap to GCC by itself+mescc-tools) <OriansJ>well slow I can fix, given time and the desire to do so <OriansJ>In retrospect janneke, I think the amount of work we did together boarders on insane and what remains just seems so small relative to it but still feels so huge for some reason. <janneke>not to mention the great gifts of nyacc and gash, i did only very little work there, and i personally wouldn't have been able to do it without guix keeping me straight <janneke>but the amazing mental support was here <janneke>had we been sane, we would have given up two years ago or so :) *janneke goes for another (6th, 7th?) round of mes-minimal-bootstrap-tarball onto tcc-boot <civodul>sanity is the enemy of great things :-) <janneke>civodul: you should know :) ... and thanks :)) <OriansJ>civodul: here I was thinking inexperience in the problem domain simply blinded us until emotional motivators helped our sane problem solving to an epic conclusion <OriansJ>janneke: you are absolutely right nyacc and gash really helped to fill some important gaps in our work. <OriansJ>janneke: also, assuming I haven't broken mes-m2 too much; we may wish to make sure the mes-m2 (built via gcc) to GCC bootstrap path is in order (as the removal of guile from the guix bootstrap binaries will be the largest reduction) <janneke>i still think it would be really cool to merge mes-m2 back into mes <OriansJ>janneke: actually I was hoping for something a little more ambitious <OriansJ>especially given once we know the gcc compiled mes-m2 works <OriansJ>which would put our bootstrap binary size at approximately 932kb by the way <janneke>yeah, what this project needs is some more ambition, even if it's just a little bit <OriansJ>janneke: well is it too much to ask after we complete the whole mes-m2 to gcc bootstrap to extend MesCC to the degree it would be able to build guile and gcc 4.7 directly?