It’s a cool day today, yet the bus driver decides to put on the air conditioning like it’s the middle of summer. Feels like I’m traveling in a fridge. 🥶

I had no idea that Google I/O was on this week. Usually this kicks off my conference watching season, but with my upcoming trip, it’s taken a bit of a backseat this year. Besides, you can only watch so many announcements from Google for things that eventually get killed six months later.

I’ve been trying to get a site up on Netlify that can serve assets from Git LFS. The only way to enable this is with the CLI tool, and whenever I try to run it, I get a “socket timeout” error. Might need to look at a simpler ways to do this, even if it means ditching using Git to deploy the site.

Had a bit of success with organising the roadmap for the squad I’m running. I decided to see if good old Obsidian could help me here. I just needed something that would present each roadmap item in a table and allow me to do some basic sorting and filtering. I probably could’ve stuck it into an Obsidian note, and used the advanced-table plugin, but I was hoping for something that worked with structured data.

So far, the db-folder plugin seems promising. What it does is read the front-matter of a bunch of files in a directory, and presents it as a table. Each column in the table is backed by a particular front-matter entry. You can add new columns, assign them types (string, number, select, date) and sort by or modify their values. Changing the cell in the table also changes the cell in the front-matter. The only downside is that each row has to be a separate file, but I think I can live with that. It might actually be a benefit.

So what I’ve done is create a separate file for each roadmap item, added front-matter entries to represent the “subcategory” (frontend vs. backend) and priority of the item, and assigned them to columns in the table. Now I can order by priority and see which part of the squad can work on each item within a single view. Pretty good.

I decided to lean a bit further into this and add the Jira JQL query for all the tickets for each roadmap item to the front-matter as well. I wrote a bit of code that’ll run this query through jira-cli and generate a nicely formatted Markdown table of the tickets, which I include in the file. It’ll also use the ticket statuses to write a one-line summary status of the road-map item (“not started”, “in progress”, “ready for release”) to the front-matter, and that is also shown in the table. The whole thing is kicked-off with a Keyboard Maestro shortcut.

This is all working pretty well. I kinda wish I can paste screenshots about the setup here, but… you know, corporate secrets and all that. Plus the whole thing is still evolving as I work out what is useful to me or not. But I this approach has got legs.

Some might say this is all procrastination to avoid dealing with Jira tickets. To that I’d say… ssh, don’t tell anyone. 🤫

It’s quite something seeing developers using Go to build (toy) operating systems. As much as I like Go, I wouldn’t call it something designed for such low-level use cases. Then again, they’ve made Go work on microcontrollers, so it’s probably easier than I think.

The state of modern office software is depressing.

I want to have something tailored for tracking the roadmap for my squad: current priorities, release dates, etc. Ideally I’d like to organise this all in some form of structured way, so that I can slice the data in different ways: producing summaries, overviews, kanban boards, etc. Because this is company data, it should live on my work laptop or on work’s systems: the cloud’s a non-starter here. It could also be great to share this with other teammates.

What are my options here? I could use Jira, but that is a shared space and very rigid in how it want’s you to define the task. That’s good for writing out the work that is ready for others to pick-up but annoying for anything higher-level than that. I could use Confluence, but Confluence is not geared towards anything resembling semi-structured data. Everything is so manual, and there’s no great way to separate data from presentation. You’re left with just adding pages with tables everywhere. It’s barbaric.

The only alternatives are all the office software that’s installed: Pages, Numbers, etc. Numbers might work: at least you’re storing data in a table. But it’s difficult to use the data in any domain other than searching and sorting (well, I guess pivot tables might help with summarising data). It’ll be difficult to present this data in any other forms, other than graphs. And sharing it with others would be nice.

What would be nice is to use some of these low-code development tools like ToolJet, or flexible databases like NocoDB or BaseRow. They can be centrally managed, configured to talk with other tools that the company is using like Jira, and available to anyone to build these small utilities or databases that you come across in your roles. Which means: no asking for IT to create these little apps.

I guess in the old days this would be the type of things that you’d get a skunks work team to build. That’s pretty heavy handed for anything that I’m thinking of. Something that can be quick to build out, say in like an afternoon or so, and that can be improved over time if others find it useful, or otherwise junked if it’s more trouble than it’s worth.

But part of me doesn’t see this happening anytime soon. And if I’m being realistic, that’s understandable. I’m probably the only person who wants this. Ah well. I guess one can dream.

Oof, someone save me. I just realised that my flight to Europe would’ve got me a day earlier than my hotel booking. Thankfully I manage to move the flight a day later. But boy, I’m so glad I checked! 😮‍💨

Now to go through every booking and make sure all the dates line up.

Is the NOT_FOUND response codes in gRPC an actual error, or could one use it to indicate a resource doesn’t exist — and it doesn’t matter if it doesn’t — on the happy path?

There seems to be a bit of disagreement to what degree non-200 OK error codes should be used to signal non-error conditions. I’m generally of the opinion that all the 4xx HTTP response codes are not actually errors, and should be available to the developer to indicate that, while the method execution was not quite as expected, it wasn’t adverse in a way that warrants error logs or PagerDuty alerts. That’s what the 5xx response codes are for.

When I discussed this with someone at work, they came up with a decent enough counter-point. Suppose you’ve accidentally missed-configured a URL of a micro-service, and when you hit that URL you get 404s. Arguably that does warrant error logs and PagerDuty alerts, especially if the quality of service is degraded and you want to get it fixed as soon as you can.

So I can see how the HTTP response codes for a REST-API can be ambiguous here. But is it the same for a gRPC API? Since gRPC has UNIMPLEMENTED to indicate a miss-configured service, can a developer use NOT_FOUND for out-of-band non-error signalling? Should error codes be used at all, and the response code be OK with the response body itself having in-band signalling to indicate that a resource isn’t found but that’s OK?

Don’t know the answer to this.

I just got news today that work has decided to move office. Somewhere with much better public transport links (i.e. a tram line). As someone who’s been on this frickin bus for 40 minutes, I couldn’t be happier.

Seen a few posts on Micro.blog over the last few days about people sleeping badly. Might be that those posts are contagious because today it’s my turn to post about how badly I slept last night. Worse was that I had to get up at 5:40 to go into the office. Ah well. Hopefully tonight. 😫

I’ve found myself using ChatGPT to do quite mundane things recently, like get the postcodes of cities around the world. It feels like I’m barely pushing the limits of what it’s capable of. I just prefer the chat-like interface it has.

And I think that’s the danger that people running “real” websites have when it comes to these chat bots. I’m sure there’re a bunch of sites out there that can get me the postcode for any world city I throw at it. But dealing with a crappy web interface filled with ads does not sound appealing to me. Much better is the clean chat interface that ChatGPT provides, complete with a history of all my prior requests that I can just look back on.

Also, and this is just my current use case, I’m not too phased with how accurate these postcodes actually are. These are all for test addresses to test a Stripe integration. So near enough is good enough.

Much of the personal stuff I’m currently doing outside of work involves writing. Lots of writing. So much writing. It’s not like a novel or anything: just a couple of websites I’ve decided to work on. Even so… still doesn’t give me the same buzz as coding does.

I’m going to be putting on my “I don’t like fun” hat for a minute and say I’m so glad there’s no “verb” for posting on Micro.blog. No tweeting, tooting, truthing, skeeting. None of that. Just good old fashioned “posting”, like the blogging gods intended. Okay. Hat off.

I’ve abandoned any further thought of self hosting my code, at least for now. As predicted, I found other shiny things to distract me away from that path to madness. I can still use custom domains for Go packages if I wanted to, so I’m not really loosing much by keeping it all in GitHub.

I’ve set up an instance of Linkding within PikaPods this afternoon. I’m going try it out as my bookmark manager, given that my Pinboard subscription has expired.

So far, so good. Importing my bookmarks from Pinboard was a breeze. The UI is cleaner that Pinboard as well, although I wish I could change the colour scheme (the purple is a but much).

Bulk editing of links was easy to do. Just run a search and click the “card” icon next to the search box. Checkboxes appear next to the links and you can select the ones you’d like delete, archive, or add or remove tags to. All pretty straightforward.

No social features though, which makes sense given that this is something you host yourself. But I never used the social features from Pinboard, so I won’t miss that at all. Poking around the settings it does look like you can make links public, but I probably won’t need that anytime soon.

One thing I’m excited about is that it sets the title of new bookmarks to that of site automatically if I don’t set one myself. Thank you! Finally! The fact that Pinboard didn’t do this, and just set the title to “No title”, really annoyed me. Such a bleeding obvious feature to add as well, and I don’t know why Pin… okay, breath, not a problem anymore. 😮‍💨

So yeah, we’ll see how this goes.