Software Engineer working at Sourcegraph building a planet-scale (readonly) IDE. Likes compilers, build tools and editor stuff. Lives in Drøbak, Norway.
@donsbot FYI, we landed a significant navigation quality improvement in scip-typescript related to “goto definition” for properties of object literals https://github.com/sourcegraph/scip-typescript/pull/373
I’ve been embarrassed about this gap to tsserver for a long time, had an open PR for almost a year but didn’t like the hacky implementation, and my colleague Christopher who joined a year ago found a much much cleaner solution.
@donsbot no ETA for 2.0. I’d love to see it supported and I’m happy to give guidance (or send out intros) if you guys want to contribute towards making it happen. We have strategic/important customers who are using the Kotlin indexer so 2.0 will likely get supported sooner rather than later.
@donsbot if you’re using the TS indexer then you might be interested in https://github.com/sourcegraph/scip-typescript/pull/256
It improves the TS navigation experience quite a lot. It should be a low-risk change to merge as well. The only reason I haven’t completed the PR is because I got pulled into Cody around that time and I’ve been working full-time on Cody since then (and it’s been kind of fun doing something new for a while ☺️)
@donsbot that’s a bummer, but maybe not surprising. Is there any appetite from your end to contribute to an updated plugin? I suspect we’ll need to support 2.0 eventually, just a matter of when.
@donsbot btw, fwiw, I experimented with adding structured signatures and types to SCIP in this PR here https://github.com/sourcegraph/scip/pull/231
Have a complementary PR to scip-typescript that we’ve been using to power TypeScript->Kotlin codegen (JSON-RPC bindings that are defined in TS and reused for IntelliJ plugin).
We also have structured signatures for Java and Scala (via SemanticDB).
@donsbot no ETA that I’m aware of. I’ll ask the team and get back to you next week once I’ve returned from PTO. You know if there many major breaking changes in 2.0 or is it mostly about bumping up versions?
Seeing many expat parents with young kids stressing over what language to speak with their children. FWIW, I grew up in a mono-lingual family and yet I’ve ended up speaking six different languages at home/school/work over the course of my life. Learning a language through real-world needs can be just as effective, if not more effective, than growing up in a multi-lingual family. Speak whatever language helps you bond with your kids the best.
After many months of having a ChatGPT subscription with the app installed on my phone (and even ChatGPT action button), I still default to Google in Safari for almost any queries on the go. Maybe time to try Perplexity.
@hipsterelectron I think image descriptions is a slightly different situation. The why behind an image posted with a tweet is usually included in the tweet itself, it’s ok if the alt only visually describes the image.
A hill I will die on is that git commit messages should explain *why* you’re making this change (and optionally why you went with your implementation strategy over alternatives). All the demos I’ve seen for AI-generated commit messages only show examples *what* is changing, which is less useful.
@hipsterelectron totally agree. The official Rust marketing is absolutely spot-on and reasonable.
@spoltier developer tooling brings me personally to Rust. On a second thought, the word “niche” is probably inappropriate for Rust is good because it excels in many domains and learning to program in Rust alone makes you a better programmer even in other languages. I just don’t see a future where Rust becomes a mainstream language for general-purpose applications like Java/TypeScript/Python are.
Programming languages I’ve primarily written this year: TypeScript, Rust, Java, and Kotlin. A bit of Python to analyze csv files, as well. Lots of tree-sitter queries too. Anticipating even more Rust in January.
I’ve learned to appreciate Kotlin enough to understand why modern Java is not yet enough. Context receivers and nullable types are powerful.
Rust is IMO oversold as a *mainstream* application programming language. It’s an excellent language for more niche applications.
Electricity stats for 2023:
- Total 17.5 kWh
- Around 4 kWh for EV charging
- Average price around $1/kWh
- Peak usage in December at 3 kWh
- Used 22% less electricity compared to average house of similar size in Norway, which I’m guessing comes mostly for “free” from having a newly built house with modern insulation
Local dog walker taking advantage of the snow on this beautiful sunny day
Scary observation about how deeply social media companies control the narrative. I see an unreasonably large number of posts about Tesla in both my Twitter and Threads feeds and they’re consistently positive about Tesla on Twitter (unsurprisingly) and negative on Threads (surprisingly?). Good reminder to be critical of where you source your opinions.
FWIW, I’ve had a Model Y for almost one year now and it’s an amazing car IMO despite my critical opinions about Elon Musk.
Clearly written case for why the Protobuf wire format is unsuitable for applications where low-overhead *write* performance is critical. I’m a big fan of Protobuf for its wide language support, binary compatibility story and good enough performance in most cases. For write-heavy latency-sensitive applications, use another wire format. https://richardstartin.github.io/posts/dont-use-protobuf-for-telemetry
@eed3si9n love myself good Kenyan beans. I’ve been getting delivered light roast beans from a roastery in Oslo for several years now and the Kenyan beans are consistently the best. I need to get more Ethiopian into my mix!
If you are doing serious work with tree-sitter then you should be using queries and Neovim’s built-in query playground (:InspectTree and :EditQuery) is really excellent.