Reports / July 2021

Until today, I’ve been working on the basis of monthly reports for Encode. There’s a reasonable track record there, with a history going back to 2016.

However, it’s often been difficult to work on these reports in a meaningful way. The rhythm feels wrong, I’m not able to recall sufficiently over such a long stretch of time, and I don’t feel like I’m imparting useful information and visibility onto my work.

Moreover, when added to the stresses of monthly accounting, invoices, and taxes, it ends up being not fun.

Recently I’ve been struggling hard to keep up with them.

So, let’s make a change.

Weeknotes: Friday 16th July, 2021.

This week I’ve switched over to writing weeknotes, I’m hoping that this’ll work much better for me that the previous monthly approach.

Some other tech folks, and their week notes…

• I’ve long admired Aymeric Augustin’s approach to his “Multiple Template Engines for Django” crowdfunded project, with it’s weekly updates. That one is from way back in 2015, but I’ve often thought of it when looking for good examples of crowdfunded accountability in open source software.
• Simon Willison writes little and often, and it seems to work well for him.
• Carlton Gibson has developed a good lil’ weekly writing routine, and I often feel like I’ve got a lot to learn from his approach to developing a healthy working balance.

For now I’ll collate my weeknotes into monthly pages, since that’s the existing structure that I’m working with.

The Encode website is a Jekyll site, hosted on GitHub pages.

HTTPCore, the directors cut.

I’ve been spending a good chunk of time doing some in-depth work based on the core networking used by httpx.

In order to provide good visibility onto my motivations and aims here, I’ve outlined the approach in a GitHub discussion.

There’s always a balance to be judged between starting fresh on something and iterating on an existing piece of work. My intuition here is that I’ve learnt enough from working on httpx and httpcore that it’s worthwhile to be drawing on a clean slate here.

Future thoughts

Some idling thoughts, worth putting out there perhaps, at least for my own recall.

• Should I be pulling MkDocs under the “Encode” umbrella?
• How can I be help maintain Django REST framework better, given that I’m spending a lot of in-depth time on HTTPX. Could the docs do with a refresh? Could we be more efficient by converting it to some of the structures and automations we’ve been learning from in other Encode projects?

Weeknotes: Friday 23rd July, 2021.

I’ve been working again on the revised HTTPCore, with a pull requests that integrates it with HTTPX.

The bits of this that I’ve been particularly pleased with:

• 100% code coverage for the test cases, excepting a couple of awkward-to-test lines that I’ve excluded. I find that having an enforced metric as part of the CI is enormously helpful. I don’t much enjoy working on project that don’t have this, except sometimes in the very earliest stages of development.
• Being able to finesse the network backends, and start to get them into shape where they can eventually be public API.

On the flip side, I’m somewhat reconsidering what the actual priorities for HTTPX ought to be right now. There’s very little reason that we shouldn’t be promoting the library to a 1.0 version right now, in order to indicate API stability. Or at least determining if that are any blockers to prevent us from doing so.

We should probably also be prioritizing getting HTTPX listed on the official Python documentation.

The rejigging of HTTPCore could be something that occurs on the other side of both those two tasks.

Weeknotes: Friday 30th July, 2021.

I’ve been drafting up a potential Kickstarter funding drive for HTTPX, which I’d expect to launch alongside the 1.0 release.

Good writing always takes far more time than I expect. I think there’s a convincing case to make for a funding drive at this point, and so I’m trying to build a cohesive story around the work that’s already gone into HTTPX, and the work that I’d like to continue doing on it.

I’ve also been working more on building HTTPCore directly into HTTPX.