James O'Gorman

@jamesog
231 Followers
252 Following
3K Posts

General geek, gadgeteer, tea drinker, occasional photographer, wannabe musician, and liker of Oxford commas. Recovering sysadmin. Lefty woke liberal. 🏳️‍🌈

I tend to post about anything, often techy things, but of late occasional UK political things too.

I mostly write code in #Go (#golang) but lately I've been learning #Swift and #SwiftUI. I will touch #Python under duress. #Perl was my first real language and made me love writing code.

(Rarely updated) Bloghttps://jamesog.net
GitHubhttps://github.com/jamesog
Flickrhttps://flickr.com/photos/jamesog
Blueskyhttps://bsky.app/profile/jamesog.net
Show threadJames O'Gorman6h ago
@kinnison Yeah, not hard, just... sigh :-)
James O'Gorman7h ago

@kinnison Just for you I've updated both my systems to 25.05 😘

Although I'm leaving stateVersion as-is, at least on one of them, because it implies an upgrade of postgres and I need to do a dance.

James O'Gorman13h ago
Queens of the Stone Age playing Royal Albert Hall, you say? Tempting...
Show threadJames O'Gorman2d ago
@kinnison For now it's a private repo on Codeberg. I've been contemplating whether it's safe to make it public, especially as I have agenix secrets stored there. In theory that should be fine, but...
James O'Gorman2d ago

Finally, all of my NixOS hosts are managed using a single flake rather than a configuration.nix on each machine. The flake also has my home-manager configurations so I can use it on my Macs too.

Took some wrangling to figure out how to package things and write a custom module to make some of it work, but Claude Sonnet 4 was surprisingly helpful there - way more than any official Nix docs.

James O'Gorman4d ago

Just had a little visitor. 🦋

I’ve seen plenty of white ones fluttering about this year, but this is the first non-white I’ve seen.

Show threadJames O'Gorman4d ago
@kpl Definitely. And thinking about idioms I might use that either don't translate or aren't known outside of this little island. It made me a better documentation writer, too.
Show threadJames O'Gorman4d ago

@kpl On:
"You’re sarcastic, and it’s not productive.” That hurt, and made me rethink how I worked."

I had very similar feedback once! It was a real wake-up call that most of my team weren't native English speakers and often couldn't tell English sarcasm meant as "humour" or when I was being serious. That made me think a lot about how I communicate with people.

Great post :-)

James O'Gorman4d ago
iFixit5d ago

What’s better than fixing something yourself? Doing it with tools that didn’t cost full price.

Now through July 17, get up to 15% off select iFixit tools. That drawer full of broken stuff just became a to-do list.

Check it out:
https://www.ifixit.com/Shop/Sale
—
#iFixit #ad #Repair #Sale #RepairTools

Show threadJames O'Gorman5d ago
@uthark Agree, although I would usually ask the submitter to squash and update the branch.
×
James O'Gorman6d ago
MFW my carefully crafted commit message gets squash-merged into a one-line commit
Show threadDaniel Silverstone6d ago
@jamesog squash merge is the bane of git forges and should be banned
Show threadJames O'Gorman6d ago

@kinnison Indeed. I am very anti-squash and disable it on all the repos it's in my power to do so.

I also wish GitHub would get around to adding ff-only merges. Merge commits are my other big pet peeve.

Show threadDaniel Silverstone5d ago
@jamesog I thought that they offered rebase/ff but not my favourite (semi linear)
Show threadJames O'Gorman5d ago
@kinnison Yes, rebase is an option, but not straight merge --ff-only. I’ve seen rebase merges go wrong and break main, where a ff merge wouldn’t have done
Show threadDaniel Silverstone5d ago
@jamesog If a rebase-merge broke then an ff-merge would have been rejected, no?
Show threadJames O'Gorman5d ago
@kinnison Correct, which is what I'd want. Most of the time rebase is fine, but once in a while GitHub says "branch has no conflicts, you can rebase" and then introduces some subtle error. FF prevents all of that by ensuring your branch must be up to date.
Show threadDaniel Silverstone5d ago
@jamesog aah I see yes that's no good. I really like the combo of ff and merge commit that gitlab's semi linear history approach gives. You can see a single line of commits, but they are grouped into logical sections by the merge commits.
Show threadDaniel Silverstone5d ago
@jamesog and to reference your rebase grump, if you have to rebase to do a semi linear merge then gitlab reruns the Ci before the merge
Show threadOleg5d ago
@jamesog squash merge is the only solution to a 50-commit PR with commit message “fix” in 49 of those.
Show threadJames O'Gorman5d ago
@uthark Agree, although I would usually ask the submitter to squash and update the branch.