I work with the financial services group in a mid sized public company, and it’s abundantly clear that writing code is not the bottleneck for any of our work. It’s knowing what code to write, maintaining multiple environments for testing and integration, each of which is composed of a half dozen vendors/systems, and making sure our audit and controls are preserved.

We have a very small amount of custom code, and we have to know it intimately. I would imagine this a common situation across many industries with any kind of controls, audits, legal or fiscal responsibilities, and av sense of professionalism.

Made good progress on my http client in C, thanks to excellent #openbsd man pages and RFC 2068. Almost done response parsing and figuring out a model for handling large response bodies.

I’m not used to doing IO at this level, and need to learn the performance characteristics of many small reads backed by the kernel buffers as opposed to a few bigger reads and managing my own buffers, like a buffered stream reader. My hunch is that the latter is preferable, fewer context switches.

Also got my email forwarding, and am learning to handle my full load with the mail client in base. It’s very capable and simple,

It all went off the rails with html email and JavaScript…