Search
Saturday, 28th of March 2020, 15:05:29 UTC
20:17:56
karlosz
there's so much work that alloc tramp duplicates with c call out
20:19:38
karlosz
do we really need to duplicate the c-call out code 4 times in each backend? i think the setup in c-call should be general enough to encapsulate everything
20:20:35
karlosz
maybe define a static lisp function called alloc tramp which does alien-funcall that allocation calls out to?
20:47:09
stassats
Krystof: do i have time to correct a NEWS entry?
20:53:17
stassats
i did it anyway, sorry if that'll make you redo something
20:54:24
Krystof
I don't know what that's just done
20:54:34
Krystof
probably I do have to redo everything
20:55:48
stassats
hope you have it highly automated
20:56:42
stassats
but you described my test case, not the change, i guess i should've been clearer in the commit message
20:57:02
stassats
that's what i always get when committing without rebuilding
20:57:56
Krystof
thanks for the heads-up
20:59:00
stassats
gotta set up a pre-commit hook one of these days
21:03:44
stassats
maybe NEWS should be edited before the freeze? so it could be debugged as well
21:04:09
Krystof
I started writing a freeze.sh script to catch all these things
21:04:18
Krystof
then I had to become a home educator
21:04:19
stassats
i didn't describe (values-list (list 1)) because it's quite boring, as we already handle (values-list (list 1 2))
21:04:37
stassats
and then it's a boring description anyway, as that's actually useful for (apply x (list 1 2))
21:04:48
Krystof
things that are boring to us are exciting to other people
21:05:03
stassats
but i guess people are clamoring for lengthy NEWS entries
21:06:09
stassats
i hope utf-8 by default would be a hoot
21:21:59
karlosz
how would i get a riscv64 binary up? adrian glaubitz asked for one
21:22:34
stassats
and following something? is there something to follow?
21:24:16
joshe
I've always taken clues from release.sh and doc/internals-notes/release-checklist.txt
21:25:01
joshe
but that doc is likely out of date in some way
21:25:33
stassats
scp sbcl-2.0.0-ppc64le-linux-binary.tar.bz2 stassats,sbcl@frs.sourceforge.net:/home/frs/project/s/sb/sbcl/sbcl/2.0.0
21:25:36
stassats
scp platform-table.html stassats,sbcl@web.sourceforge.net:/home/groups/s/sb/sbcl/htdocs
21:25:47
stassats
after make generate-pages
21:25:55
stassats
and editing platform-support-platforms.lisp
21:26:40
karlosz
okay, thanks, i've stashed that
21:28:49
stassats
my whole cheat sheet https://gist.github.com/stassats/d56197c3499d6cde2b8111abef08bb6f
21:29:11
stassats
to get the right directory names etc.
21:30:58
Krystof
I think there's a "make upload" in sbcl-page
21:31:29
stassats
scping a single file is faster
21:34:30
stassats
one of these days i should make auto-releases on tags on github, and script binary generation for arm, ppc through the gcc compiler farm
21:34:49
stassats
but it's a really painful job
21:45:46
karlosz
yeah automating it would be great
21:46:20
karlosz
although i'm stuck here porting to platforms i have to emulate on with images i need to compile myself
21:48:09
karlosz
though riscv32 works now with only +10 insertions -6 deletions so that's two platforms to put binaries up for
21:48:16
karlosz
do we build with contribs in the binaries?
22:05:27
karlosz
i tried to push to sbcl-page just now but i get fatal: remote error: access denied or repository not exported: /p/sbcl/sbcl-page
22:06:29
karlosz
although it didn't prompt for username so that might be a problem on my end
22:07:25
karlosz
okay i had to change git:// to https:// and it worked
22:07:38
stassats
i have ssh://stassats@git.code.sf.net/p/sbcl/sbcl-page
22:07:57
karlosz
my mails are still sending as apache---
22:08:48
karlosz
i don't think i'm an http server but i could be wrong about that
22:09:32
stassats
a helicopter? some native american ancestry?
22:13:29
karlosz
now how am i going to make cross build runner happy about the fact that i want to check with both -32 bit and +32 bit
22:13:43
karlosz
or rather -64-bit and +64-bit
Sunday, 29th of March 2020, 3:05:29 UTC