#pest | 2022-05-16

04:56

signpost

shinohai: http://trinque.org/src/pentacle/ << rsync'd all current builds and v trees, including clisp and sbcl

↖

04:58

signpost

sorry it took so long, just met some new neighbors, nice folks.

04:58

*

signpost back tomorrow

~ 8 hours 7 minutes ~

13:06

shinohai

http://logs.bitdash.io/pest/2022-05-16#1005474 << Excellent and thanks again m8.

13:06

bitbot

Logged on 2022-05-16 04:56:45 signpost[asciilifeform]: shinohai: http://trinque.org/src/pentacle/ << rsync'd all current builds and v trees, including clisp and sbcl

~ 42 minutes ~

13:49

shinohai

$ticker btc usd

13:49

busybot

Current BTC price in USD: $29597.11

13:57

shinohai

clisp failing to build for me with this barf: http://btc.info.gf/paste/d60007@raw

14:07

shinohai

Tried going into tmp/ build directory and setting executable bit on `install-sh` but *then* barfs with "unix.d:139:56: fatal error: asm/page.h: No such file or directory"

~ 3 hours 14 minutes ~

17:22

signpost

hmmm

17:23

signpost

yeah, I cut those patches wrong somehow. sec

17:36

signpost

ok, rebuilding over here and then I'll reupload clisp's second patch.

17:36

signpost

and the pbuild

17:37

signpost

shinohai: this is making me want an on-disk vpatch-friendly bug tracker.

17:37

signpost

clisp's makefiles have a parallel-build bug for example, looks like.

17:38

shinohai

That would be a nifty thing to have, so wouldn't hafta sift through logs for bugs.

~ 32 minutes ~

18:11

signpost

looks like I was relying on my copy of clisp src in /src/work/clisp. there are binary translation files it relies upon to build that don't make it through vdiff.

18:11

signpost

so what I patched wasn't representative of what I was building.

18:11

signpost

will get at ya when I finish up.

~ 1 hours 7 minutes ~

19:19

signpost

alrighty, re-uploaded to http://trinque.org/src/pentacle

19:19

signpost

works over here with the workdir moved out of the way, and sbcl build worked atop it

19:24

signpost

had to add a step to the build to generate the gmo files from po with msgfmt, and remove references to a png in documentation build.

19:25

shinohai

kk thanks, will fire it off & report back

19:25

signpost

sg, ty

19:28

signpost

in other news, I've got the gc build within a few bytes in cc1 and cc1plus of reproducibilty.

19:29

shinohai

winning

19:32

shinohai

heh so was I doing it wrong by trying to build clisp by itself?

19:33

signpost

you can, but shouldn't have to, as sbcl declares it as a dependency

19:33

signpost

clisp's really only in there to fart up an sbcl.

19:34

signpost

btw sbcl's not currently statically linked, will require presence of the musltronic libc.so

19:34

signpost

there were particulars regarding how sbcl uses dynamic loading for fasls I haven't ventured into yet.

19:36

signpost

https://www.timmons.dev/posts/static-executables-with-sbcl-v2.html << this guy's working on patches apparently, haven't tried them.

19:36

shinohai

lemme try doing sbcl without clisp first, now I'm hitting this error when trying to build clisp alone http://btc.info.gf/paste/cc679f@raw

19:37

signpost

sbcl will not build without another lisp present with which to bootstrap

19:37

signpost

ah you have an old pbuild, sec

19:40

signpost

http://trinque.org/src/pentacle/pbuild

19:40

signpost

I added quoting around those macro definitions

19:41

*

shinohai also plans to mirror pentacle src tree soon ....

19:42

shinohai

Seems to be chugging along fine with new pbuild

19:43

signpost

ossum

19:43

signpost

my goal for today is to get pbuild itself as vpatches.

19:51

shinohai

clisp great success!

19:51

shinohai

now for sbcl

~ 16 minutes ~

20:08

signpost

aaand I deleted my /dev again by not unmounted before trashing a chroot, lol

20:08

*

signpost reboots

~ 28 minutes ~

20:36

shinohai

signpost: has the zlib pbuild changed? sbcl fails with `coreparse.c:48:19: fatal error: zlib.h: No such file or directory` though zlib installed.

20:38

shinohai

hmmm quick vdiff shows mine differs from what's in current tree, imma reinstall

20:39

signpost

yeah, current sbcl pbuild oughta say require clisp and require zlib at top

20:40

signpost

http://paste.deedbot.org/?id=5Jwg

20:42

shinohai

Had zlib installed already but dunno why it couldn't find, retrying now.

20:55

shinohai

That seems to have squared it away, updated zlib has sbcl building now.

~ 19 minutes ~

21:15

signpost

turns out clisp's translation files are usually generated by... (drumroll) an existing clisp binary, though was able to substitute msgfmt from gettext-tiny.

21:15

*

signpost hypothesizes a psychological reason for the urge to do this, in addition to w/e convenience by getting to write lisp and not lower-level lang.

21:16

signpost

something like the belief that there can be loops in causation.

21:16

signpost

self-caused objects.

21:17

signpost

there can't; the thing's caused by a long chain of previous versions of clisp. but I think there's something existentially satisfying about the fiction that it's self-caused.

21:19

signpost

there's something of an essential delusion of man in this, maaaan.

~ 38 minutes ~

21:58

asciilifeform

signpost: the need for a sane bootstrapper for lisps still gaping, aha

~ 55 minutes ~

22:53

shinohai

Well got further this time, stopped here http://btc.info.gf/paste/900bfb@raw

22:53