Rest in peace, Tomáš :-(
https://prl-prg.github.io/tomas-kalibera.html
Remote
CF Bolz-Tereick
@cfbolz@mastodon.social
mastodon 4.8.0-nightly.2026-10-06PyPy/RPython contributor. Half time teaching at Uni Düsseldorf. Works on dynamic language implementations. Vegan. Love street art and art in public spaces, hiking, reading.
they/them
0 Followers
0 Following
33 Posts
Joined April 08, 2017
website:
Yay, @stanfromireland@mastodon.social just became a PyPy core dev and merged the first PR! Congratulations and welcome! https://github.com/pypy/pypy/pull/5424
Together with Oskar Haarklou Veileborg he's the second newly active dev on the project in the last year.
Replying to
It's very amusing to me that there were two different new meta tracing jits announced this week, yk by @ltratt@mastodon.social and @verijit@mastodon.social
And both are using a Mandelbrot demo! @CanLehmann@mastodon.social
Replying to
@CanLehmann@mastodon.social @smarr@mastodon.acm.org @lindsey@recurse.social @verijit@mastodon.social verijit is actually doing the second futamura projection! The result (a residual cpu, specialized to a concrete program) is the same, but the compilation process itself is faster. (Graal is doing the first, rpython too). And yes, partial evaluation and meta tracing are not exactly the same, but have a lot of similarities (there's a chapter about the similarities/differences in my diss)
yesterday I found out that XaoS nicely runs in the browser: https://xaos-project.github.io/
I played *a lot* with it when I was like 15 and had just installed Linux for the first time. And the aa terminal mode was just the best thing ever.
Replying to
@regehr I love this post, and I hate how much nonsense discourse there is on hn about it 🙄 (I know, what did I expect)
Replying to
@smarr@mastodon.acm.org @CanLehmann@mastodon.social @ltratt@mastodon.social @verijit@mastodon.social I haven't actually decided what to talk about at icooolps. It might be this, or I'm thinking of doing a talk on pragmatic approaches for improving compiler correctness
Replying to
@regehr Easter Orthodox Christian fasting can include not eating meat, dairy, eggs. So it's not totally surprising that there are vegan dishes from the Balkans.
(it's not part of the Balkans, but I found amazing vegan food in Romania on my last visits)
'Wearing Flip Flops' is really an excellent title for a hardware podcast https://wearing-flip-flops.com/
(didn't listen to it yet though)
Replying to
@shriramk omg, what an amazing picture, thanks for sharing it! Put it on Wikimedia Commons?
Replying to
Last year there was also the excellent post 'The tech interview is a legible, reasonably well-designed process' https://herecomesthemoon.net/2025/06/legible-well-designed-processes/
Replying to
@lritter@mastodon.gamedev.place yeah it feels like it works by accident on mobile, mouse works fine and the browser turns touch events into some random mouse thing
Replying to
@lritter@mastodon.gamedev.place on the laptop it just works nicely. On the phone it works for me in full-screen mode, tap and hold to zoom.
Replying to
@penguin42@mastodon.org.uk yes, we meta trace the core. It's generic, so it works for other cores too (but we haven't tried that for a super wide variety of cores yet)
Replying to
@ltratt@mastodon.social @verijit@mastodon.social @CanLehmann@mastodon.social yeah, making any kind of perf demo is hard. 'it's literally the same, just a bit faster!' just doesn't pack that much punch
Replying to
@tom_verbeure @CanLehmann @verijit it is a generic simulator. However, the optimizations are quite geared towards CPUs. So for other designs it's less clear that it'll be faster than other simulators.
What kind of Verilog designs are you interested in?
Replying to
@penguin42@mastodon.org.uk the pycket one? That was fun. the very first meta tracing paper is this one: https://stups.hhu-hosting.de/downloads/pdf/BoCuFiRi09_246.pdf
Feel free to ping if you have questions!
Replying to
@penguin42@mastodon.org.uk @CanLehmann@mastodon.social @verijit@mastodon.social it's 'meta' because the tracing jit is parametrized by the core. You can trace a riscv program according to the Verilog code of hazard3, or some other riscv core (or do the same thing for another isa and corresponding core entirely). The jit infrastructure is shared for all these cases
Replying to
Replying to
@DRMacIver@mastodon.social this is so cool! and, incidentally, for one of my current projects the C++ version of Hegel can't come soon enough...
Replying to
@ambv@mastodon.social love the podcast! The content is great, but my favorite part is the cover of the Monty Python Flying Circus theme musing in the outro 🎶🦶
Replying to
@ambv@mastodon.social wow, seriously?!? that's a little bit disappointing somehow (I noticed it a few episodes ago). also, you should invite guests (like me) sometime ;-)
Replying to
@penguin42@mastodon.org.uk @CanLehmann@mastodon.social @verijit@mastodon.social does that make any sense? Happy to answer more questions about it!
Replying to