Neal Wilson ✅ @neal@social.typica.us

Just about finished with the chunk of code I've been working on lately. Today I worked out a new feature. Not sure if it's how I'll keep things, but I'll live with it as it is for a while and see how that goes.

Old cat is acting like an old cat.

On reflection, this was the high point of the work day.

I saw four good dogs on the way to work today.

The strange bug involved data being reported from the future, but I had just set up one of the transformation pipelines wrong. No actual precognition was observed.

Continued with the work from the past couple days and noticed and fixed a couple bugs (one strange, the other significantly less strange). Knocked a couple items off the to do list for the part of the code base I'm working on, but still have a few more things to add.

There's a bit more work to do on today's code to get all the features I want in, but the fundamentals are there with a lot less debugging than I usually need when writing this sort of code. I still hate doing graphics code, but maybe I'm getting a little better at it.

Today I'm doing graphics code. I hate doing graphics code but it needs doing so I'm doing it.

Documentation: "Don't use this, use these other things instead." Me: *looks at the recommended things, sees that they don't have the argument I need* Me: *Uses the thing anyway* Program: *works fine*

Ended up doing one more feature before bed. It's a little hacky and I might need to rework it later if I want to add more features, but that's a problem for future me. For now it's working as intended.

Had to take a brief break to unmelt my brain (not because I was trying to do something difficult but because I was about to do the easy thing in a way that was more difficult than it needed to be) but got done more than I planned on. That might be it for today's productivity, but I'll give some thought to what should come next.

Dinner has been eaten. Now to see if I can finish up another chunk of work.

Found a documentation bug, but taking a different approach entirely cut the work my code needs to do in half.

Last night I finished my latest bottle of whiskey and it occurs to me that there's most of an open bottle that came with the house, abandoned by the previous owners. I suspect it was first opened a few decades ago and I assume it's no longer delicious.

I don't need another project, but my sister is talking like I might need another project, mainly because the thing she wants I know she'll hate anything she can just buy and I won't be able to fix it for her, whereas she /might/ hate a thing I make less and fixes would be possible.

Updated the shop's web site with the new coffees today.

Testing the past couple day's code by running a sweep through the relevant range of values and graphing the output in a spreadsheet pointed me at the existence of a bug which, once fixed, validated my hunch from several years back that the extra math involved in the new code did not produce a significant difference from computationally cheaper approach I went with originally. I still want to try a blend of the two methods with live data to see if the slight difference feels better.

Had some administrative responsibilities at the shop to take care of but now I can spend the rest of the day at home coding with the cat. I should be able to slam out the idea I noted yesterday as most of what I need for that is already done.

Today's code is mostly directly pulled from 1.9.1 but I've taken the opportunity to make some minor efficiency improvements.

The idea may also be applicable to CRUCS, so I might add it there some day, though that's a low priority for me.