Banging my head against the wall trying to debug memory usage in a hybrid Rust / Python (using PyO3) program. Problem is, it only OOMs after steadily using the expected 150gb of RAM for 10 hours. Runs too slowly under Massif, so I’m not sure how to debug besides trying to correlate results from small local runs to the behavior of the longer runs…
Trying to release v1 of my roguelike. Finishing up studies. Reviewing TGGW and attempting to help with updating the manual, etc. Have to complete my part of the 7DRL judging process. Also probably moving on the weekends for good measure…
Going to try to finish up my port of plan9port’s devdraw(1) to Wayland. It’s usable day-to-day (pointer warping is back!), it’s just a matter of figuring out how long I want to try to shake bugs out of it.
Putting the final bows and ribbons on the Python bindings for starlark-go that I’ve been working on, and then getting back to working on the tool I wanted the Starlark bindings for.
We just hired another developer (!!!), so this week and last are a bit slower while they get up to speed. Other than that, we had a semi-successful launch that yielded even more work to do. Anyone else that has done startups notice that tipping point where it goes from making your work to having the work made for you? Exciting, stressful, and different all at the same time.
That’s a dangerous period as a dev at a startup. You really start to risk losing sight of the actual project/product goals in favor of chasing down features that your clients mentioned in a sales pitch meeting.. Pushing back as engineering can become one of the most important parts of your job.
I can second that. I’ve worked in a tech company that was hired by a startup to build their product, and at some point they started asking us to implement every feature that ever got mentioned by potential customers. This caused them to lose sight of the strengths of the product and re-target to a new niche market every week. In the end their product was a watered-down mess and good at nothing in particular, causing their original niches to abandon the product too.
Since we were “just the implementers” we never really provided a lot of pushback, but the constant massive changes required for these new features (which went pretty deep) were not good for the code quality and turnaround time, either.
Yep agreed, a healthy back and forth between the devs and business needs to exist since day one.
I suppose my comment was more referring to the “critical mass” that an organization reaches after a certain point. Now, based on the size of the team and customer requirements, work is being driven extrinsically rather than intrinsically. Not mutually exclusive by any means, but now work just… is?
Oh yeah, that definitely happens. That part, in and of itself, is actually a very good feeling I think. You no longer have to convince anyone that your work is needed, it’s self-perpetuating.
First task is finally giving in and converting our dev env from Vagrant to Docker because the latest macOS security update seems to have broken something obscure to do with multiple fs mounts on Virtualbox on Big Sur/x86, and it turns out not everyone likes fiddling with unix shells all day like me. It’s also prep for when Apple’s reported “production line issues” (read: chip shortage and China in lockdown) finally relent enough for my new M1 laptop to get assembled and delivered months later than originally promised.
I rewrote and published a tool I built to list possible solar PV arrays from a set of electrical constraints and PV modules: https://pvsizer.davis-hansson.com/
First time building a real thing with Typescript and Vue! I’m not surprised to find I like Vuejs.. but I have to admit I am surprised how much I enjoy Typescript. I’m consistently surprised at how “deep” into the code the IDE will give correct type hints.. very cool.
Who the fuck knows?!!
I’m talking at the first Cambridge RISC-V meetup later this week, so doing some work to prepare that.
On the WebAssembly side, I’m about to post an RFC on the approach to represent table types in C/C++ (as was requested on my current draft patch.
Banging my head against the wall trying to debug memory usage in a hybrid Rust / Python (using PyO3) program. Problem is, it only OOMs after steadily using the expected 150gb of RAM for 10 hours. Runs too slowly under Massif, so I’m not sure how to debug besides trying to correlate results from small local runs to the behavior of the longer runs…
Trying to release v1 of my roguelike. Finishing up studies. Reviewing TGGW and attempting to help with updating the manual, etc. Have to complete my part of the 7DRL judging process. Also probably moving on the weekends for good measure…
Going to try to finish up my port of plan9port’s
devdraw(1)to Wayland. It’s usable day-to-day (pointer warping is back!), it’s just a matter of figuring out how long I want to try to shake bugs out of it.Move to a new apartment in Amsterdam; celebrate my first King’s Day. Maybe go to a live show.
Oh, and work, I guess.
I have to port some VBA in an Access database to another language for work. I haven’t touched VBA for like 12 years.
Putting the final bows and ribbons on the Python bindings for starlark-go that I’ve been working on, and then getting back to working on the tool I wanted the Starlark bindings for.
We just hired another developer (!!!), so this week and last are a bit slower while they get up to speed. Other than that, we had a semi-successful launch that yielded even more work to do. Anyone else that has done startups notice that tipping point where it goes from making your work to having the work made for you? Exciting, stressful, and different all at the same time.
That’s a dangerous period as a dev at a startup. You really start to risk losing sight of the actual project/product goals in favor of chasing down features that your clients mentioned in a sales pitch meeting.. Pushing back as engineering can become one of the most important parts of your job.
I can second that. I’ve worked in a tech company that was hired by a startup to build their product, and at some point they started asking us to implement every feature that ever got mentioned by potential customers. This caused them to lose sight of the strengths of the product and re-target to a new niche market every week. In the end their product was a watered-down mess and good at nothing in particular, causing their original niches to abandon the product too.
Since we were “just the implementers” we never really provided a lot of pushback, but the constant massive changes required for these new features (which went pretty deep) were not good for the code quality and turnaround time, either.
Yep agreed, a healthy back and forth between the devs and business needs to exist since day one.
I suppose my comment was more referring to the “critical mass” that an organization reaches after a certain point. Now, based on the size of the team and customer requirements, work is being driven extrinsically rather than intrinsically. Not mutually exclusive by any means, but now work just… is?
Oh yeah, that definitely happens. That part, in and of itself, is actually a very good feeling I think. You no longer have to convince anyone that your work is needed, it’s self-perpetuating.
At work I’m churning though bugs, but may get pulled into a project upgrading a large library we use.
At home I’m ostensibly learning Rust by implementing a 3D snake game. So far it’s not winning me over, but we’ll see how the project goes.
Configuring my new Macbook @ work, required for working on iOS/React Native. I already hate every aspect of it.
First task is finally giving in and converting our dev env from Vagrant to Docker because the latest macOS security update seems to have broken something obscure to do with multiple fs mounts on Virtualbox on Big Sur/x86, and it turns out not everyone likes fiddling with unix shells all day like me. It’s also prep for when Apple’s reported “production line issues” (read: chip shortage and China in lockdown) finally relent enough for my new M1 laptop to get assembled and delivered months later than originally promised.
I rewrote and published a tool I built to list possible solar PV arrays from a set of electrical constraints and PV modules: https://pvsizer.davis-hansson.com/
First time building a real thing with Typescript and Vue! I’m not surprised to find I like Vuejs.. but I have to admit I am surprised how much I enjoy Typescript. I’m consistently surprised at how “deep” into the code the IDE will give correct type hints.. very cool.
Looking at a Go programs JSON output to write a different Go program to generate that JSON because I need to learn the ins and outs of this system :(