libera/#mezzano - IRC Chatlog

14:54:52 Vinapocalypse iska, I'll try that out (I wanted to last night but 15 hour workday q_q) - i saw there is a ticket about (format) having a few issues with digit padding it seems, which is why I was curious about testing

15:59:49 iska I know the printer implementation isn't the best one

16:26:57 Vinapocalypse Was Mezzano's CL adapted from another open source CL or written from scratch?

16:27:33 ebrasca Vinapocalypse: mostly froggey wrote it

16:32:40 Vinapocalypse Wow, that's a lot of work

16:32:50 iska as with many CLs some components (like CLOS) were borrowed, like xp printer and closette

16:36:11 iska wouldn't be so many CL implementations without this secret technique :p

16:37:27 Vinapocalypse lol

16:38:09 froggey hrmph.

16:38:13 Vinapocalypse feels kind of like Linux distros: "This one has a problem, better fork a new one"

16:39:05 Vinapocalypse Altho I know Linuxes and I guess CLs are usually written for certain domains

16:42:18 froggey I wrote most of it. Some small components were copied. closette was the original foundation for the clos implementation but I made a huge number of changes to improve performance & conformance

16:42:19 iska they still need to work on the low level details like the GC and compiler, so they're more comparable to BSDs

16:42:54 iska good job froggey

16:50:52 Vinapocalypse That's impressive

17:00:43 iska a compiler and scheduler would make a world of difference, that's what I'm working on right now

17:02:04 iska mezzano performs poorly with 32 and especially 64 threads

17:02:41 iska I bought a dual socket system just to optimize for this

17:08:30 Vinapocalypse :o

17:17:02 iska also housing an optane in there, been obsessed with single-level store for as long as lisp

17:17:02 iska Mezzano does run on that principle, the filesystem is a hashed tree of objects in memory.

17:18:09 froggey I'm glad someone understands that :)

17:24:54 iska also, despite imperfections the compiler uses an SSA backend, so it can be faster than SBCL if I put in the work

17:26:59 fitzsim iska: 115200 change merged, thank you

17:28:50 iska ssa is pretty much what every modern compiler uses except Scheme, sbcl's compiler is older than that and only recently got real type inference

17:28:59 iska np!

17:31:01 ebrasca froggey: Do I rejoin my fork and push things to you?

17:32:11 froggey yes, the new compiler was written with ssa & optimisations in mind. I never got around to writing many optimisation passes for it though

17:33:47 froggey the register allocator is a gargantuan nightmare thought. it's slow and it's quite easy to get it to be pretty stupid

17:52:24 ebrasca froggey: ???

17:53:11 froggey ebrasca: hmm?

17:54:18 ebrasca froggey: Do I rejoin my Mezzano fork with your Mezzano and push things to you?

17:54:36 froggey I've no idea. not my problem anymore. I've retired from development, remember

17:54:55 fitzsim ebrasca: open pull requests for individual stuff you want merged and I will look at it

17:55:20 ebrasca fitzsim: Why you?

17:55:43 fitzsim not sure

17:56:29 fitzsim I can test general patches, at least

17:56:44 fitzsim I may not have matching hardware if the patches are hardware-specific

17:57:27 ebrasca fitzsim: How do I rejoin my fork back? Why I also don't get the same treatment as you? (I did contribute a good amount of code)

17:58:24 fitzsim I don't understand what you mean by "rejoin"

17:59:06 fitzsim do you mean, you have some changes you want to get into github.com/froggey/Mezzano?

17:59:59 fitzsim if so, the best way is what I described

18:00:39 ebrasca https://github.com/ebrasca/Mezzano , because of what I did understand from froggey I did separate my fork from froggey one to don't give him work and now PR are still acepted there

18:02:10 fitzsim ok, it looks like you have worked on the TCP implementation there; I remember you trying to make it more standards-compliant

18:03:33 Vinapocalypse merge request?

18:04:27 fitzsim have to drop, will probably be back later today

18:12:08 iska ebrasca: I can help with that

18:12:18 iska I have my own local commits

18:13:32 iska what you do is copy your modified files, checkout the last upstream commit, make a branch, copy changed files back and make commit(s). Then push that to github

18:25:36 Vinapocalypse `git stash` to move the changes aside, `git pull`, `git checkout -b YOUR_BRANCH`, `git stash --pop`, then commit

18:27:43 iska stashing only deals with uncommited stuff, he has made commits to master

18:29:46 Vinapocalypse ahh yeah

18:43:50 ebrasca Sorry for the problem