ben_pfaff

I spent over a decade posting so much to comp.lang.c. I've run into a few regulars there in my work since then. I suppose it's still ticking along but I haven't visited in almost 20 years.

I use Emacs.

I decided to try using proportional fonts for coding starting a year or two back. It worked out well and I stuck with it, because proportional text is easier for me to read on the whole, and because it allowed more characters to fit comfortably on each line on average. I did find after a while that occasionally the lack of alignment between characters on two subsequent lines was a problem, but then I configured my editor so that it showed comments and text strings in a monospace font and that fixed the problem for me.

Bjarne Stroustrup wrote a paper about adding overloading for the "whitespace operator" in C++, but in his case it was a joke: https://stroustrup.com/whitespace98.pdf

> Main downside I found when testing Feldera is that their SQL dialect has a bunch of limitations inherited from Apache Calcite

At Feldera, we're adding features to our SQL over time, by contributing them upstream to Calcite, making it better for everyone. Mihai Budiu, who is the author of the Feldera SQL compiler, is a Calcite committer.

I'm new to CI auto-fixes. My early experience with it is mixed. I find it annoying that it touches my code at all, but it does sometimes allow a PR to get further through the CI system to produce more useful feedback later on. And then a lot of the time I end up force-pushing a branch that is revised in other ways, in which case I fold in whatever the CI auto-fix did, either by squashing it in or by applying it in some other way.

(Most of the time, the auto-fix is just running "cargo fmt".)

(Feldera co-founder here.) There are some cases where Feldera needs to index data indefinitely, yes. For those cases, Feldera can put those indexes on storage rather than keeping them entirely in RAM.

In a lot of cases where one might initially think that data needs to stay around indefinitely, people actually want the results from the last hour or day or month, etc. For those cases, Feldera supports a concept called "lateness" that allows it to drop older data: https://docs.feldera.com/sql/streaming/#lateness-expressions.

These days, it is absolutely unheard of for someone to get a doctorate at age 21. I don't know whether it was unusual then.

I bet you could build something like this as a pair of earrings.

I wasn't sure anyone but the authors remembered C Unleashed! I wrote the chapter on binary search trees and balanced trees.

Comp.lang.c was important to me for many years. I've met 5 or so of the regulars at least once. The most famous comp.lang.c regular is probably Tim Hockin of the Kubernetes project.

There's an indirect relationship between Rego and DDlog, at least people-wise. OPA comes from Styra, which was founded by Tim Hinrichs and Teemu Koponen. Teemu designed nlog, which is also a Datalog variant, for use in NVP, which was the network virtualization product at Nicira and was later renamed NSX after VMware acquired Nicira. Tim also worked with Teemu (and me) on NSX at VMware. And Teemu was one of the architects of OVN (the biggest open source application for DDlog), with Tim also having some influence. And Teemu also knows Leonid and Mihai (the principal authors of DDlog).

Some of the episodes of my OVS Orbit podcast may be relevant:

* Episode 5: nlog, with Teemu Koponen from Styra and Yusheng Wang from VMware (https://ovsorbit.org/#e5)

* Episode 44: Cocoon-2, with Leonid Ryzhyk from VMware Research (https://ovsorbit.org/#e44)

* Episode 58: Toward Leaner, Faster ovn-northd, with Leonid Ryzhyk from VMware Research Group (https://ovsorbit.org/#e58)

I don't know anything about Souffle so I can't directly answer the question. But there is some related material.

The DDlog repo includes a Souffle-to-DDlog converter: https://github.com/vmware/differential-datalog/blob/v0.38.0/...

The Souffle converter is somewhat incomplete: https://github.com/vmware/differential-datalog/issues/174

If you're talking about building the DDlog-to-Rust compiler, then it's not that hard. Really it's just a matter of installing Haskell then typing "stack build". But each release also comes with binaries for GNU/Linux, OS X, and Windows, e.g. for 0.38.0: https://github.com/vmware/differential-datalog/releases/tag/...

Speaking with my VMware hat on, I don't know of use cases there.

Yes, and I'm really impressed with what we've been able to do with it in OVN lately. The patches aren't posted yet (probably Monday) but (with Leonid Ryzhyk's help) I managed to take one benchmark from about 110 seconds in the C version of OVN to about 35 seconds with DDlog.

We have other internal-to-VMware use cases for DDlog. The OVN use case is the one that's easiest to evangelize since it's all open source.

I just released what I think is a well crafted e-book for a favorite old science fiction book, with permission from the author: https://twirb.github.io

I like this naming philosophy too. I once named a feature "VLAN splinters" to discourage people from using it; after all, who likes splinters? Unfortunately, people did use it and I'm only now getting to the point where I think it can be removed. I suspect that naming is, in the end, not very effective, but it still feels good.