IRC channel logs

<dannym>janneke: Wasn't at gitlab yet, no

<dannym>janneke: Good night :)

<janneke>dannym: https://gitlab.com/janneke/tinycc/-/tree/mes-0.23

<janneke>good night! (really!)

<dannym>janneke: Tried current mes-wip on ARM--all tests except two still pass. The two are the ones with unsigned "char". (7r-sign-extend, 7l-struct-any-size-array-simple)

<dannym>janneke: That's because the fix from master has not been merged yet

<dannym>janneke: Should I merge the fix from mes-master?

<janneke>dannym: yes, sure

<dannym>Rebase?

<dannym>Or on top?

<janneke>rebase and reset is fine

<janneke>ah, on top will probably also do -- it will reorder after updating master again

<janneke>i'm a great fan of keeping history linear

<dannym>janneke: It will? With git merge? It detects the merge commit?

<janneke>dannym: i never merge, just rebase

<dannym>Ok I'll rebase then

<janneke>if order is important, then reorder

<janneke>yeah, thanks

<dannym>rebased

<dannym>push: remote rejected

<janneke>i'm using git bisect heavily -- and we're small enogh not to merge, i think

<dannym>push -f: remote rejected

<janneke>yeah, to push to savannah you have to remove it first; push :wip

<janneke>(that's another reason why i push most wip stuff to gitlab first)

<dannym>Uhh

<dannym>Yeah, that sucks

<dannym>I thought it was a guix thing, but all of savannah does that? Weird...

<dannym>So I'll do "git push --delete origin wip" and then "git push" ?

<janneke>oh, i haven't really looked into that or asked -- possibly it could be changed?

<janneke>possibly; i alway do: git push origin :wip; git push origin wip

*janneke looks-up --delete

<janneke>ah right

<dannym>janneke: Well, github has no problem with rewriting history. I guess if you want to have a permanent record of commits to master, it's good to disable it I guess. But doing so for feature branches is ridiculous

<janneke>All listed refs are deleted from the remote repository. This is the same as prefixing all refs with a colon.

<dannym>Ok, pushed to wip

<janneke>thanks!

<janneke>dannym: well, yeah but otoh using a shared repo you have to be a bit careful with --delete ;)

<janneke>dannym: i tried "guix pull" on banana but ended up stealing your latest pull by copying guix current symlinks

<janneke>qemu segfaults and well, could be anything

<bauen1>OriansJ`: would https://github.com/bauen1/mescc-tools-seed/commit/ff04d45cd7f49241c6249870a8200768d99017ac also be of interest ?

<bauen1>one one side it reduces the amount of hand counting necessary by removing jump instructions

<bauen1>but probably increases complexity a bit

<bauen1>if you're interested, i'll take care of applying the change to the M1 and hex0 version

<bauen1>actually no, cmp is a signed compare ...

<bauen1>fixed: https://github.com/bauen1/mescc-tools-seed/commit/26883fbe58fa3b473176a7e5c9e0255e8026f9d2

***terpri__ is now known as terpri

***stikonas_ is now known as stikonas

***terpri__ is now known as terpri

<janneke>dannym_: well, i found why the arm-unknown-linux-gnueabihf-gcc -built tcc, i.e. ./arm-unknown-linux-gnueabihf-tcc segfaults in crt1.c; it needs -marm

<janneke>pushed a squash commit to tinycc

***awilfox_ is now known as awilfox

<OriansJ`>bauen1: reasonable modification but a little tip to remove another jmp and save some more bytes https://github.com/oriansj/stage0/blob/master/stage1/stage1_assembler-1.s#L181

IRC channel logs

2020-08-24.log