samhh.com
Human of a few decades. Loves animals and functional programming and pastel colours.
110 posts
98 followers
290 following
Regular Contributor
Active Commenter
comment in response to
post
Golden middle fallacy-ish.
comment in response to
post
Custom operators in Haskell. Very elegant at times but makes every codebase look different.
comment in response to
post
It’s the logical end result of capitalism where slop that barely works and is a poor experience is more cost effective than something better. The market generally creates a race to the bottom. This is all visible all throughout modern society.
comment in response to
post
Speaking from experience you can blame that on GitHub’s infra.
comment in response to
post
They’ve wanted this for a long time. My read is that piggybacking on migration is the politically easiest route to getting digital ID over the line, hence the argument in there not really making much sense.
comment in response to
post
Long term this will reduce innovation both in the large (PLs) and the small (patterns).
comment in response to
post
Have to be careful that the important things don’t atrophy in your brain, though.
That can also happen if you just don’t care for too long, which I’m learning as I fall out of love with programming as a hobby and treat it more strictly as work.
comment in response to
post
Just tested in a real dev scenario on Chrome and it works, but it's a bit unintuitive initially. Devtools a bit buggy with the fallback property in particular.
comment in response to
post
Works for me in the examples here: developer.mozilla.org/en-US/docs/W...
comment in response to
post
Writing out my wants I suppose Nushell comes closest but it's still so immature and poorly documented. At least I'm confident my Fish scripts will still work in a year.
comment in response to
post
Fully appreciate everyone is participating in good faith though 🙂
comment in response to
post
I think I’m still a bit miffed about pipelines struggling to get through, including the basis on which unary pipelines were rejected pre-Hack style. If memory serves the primary objections came from engine implementers, which never made much sense to me given the prevalence of userland pipe fns.
comment in response to
post
Then seeing this reach stage 3 feels like a case of bias (in the neutral sense of the word) in committee and implementors. I can't remember the last time I interacted with an API that required manual cleanup, except maybe rarely removing event listeners when React components unmount?
comment in response to
post
In hindsight this is a pretty negative message.
It's just a reflection of my disappointment in which specific proposals are failing to get through TC39. What would be relevant to me in industry are things like pipelines (😬), records and tuples, enums (as sums with data), etc.
comment in response to
post
Of all the proposals that could get through TC39 for new syntax, this one is probably (?) the least relevant to most JS devs.
comment in response to
post
github.com/samhh/dotfil...
P.S. shell is not fun to write.
comment in response to
post
Speaking of which, I’ve adopted the “tug” nomenclature for quickly advancing a bookmark.
Does the community have a preferred word to describe commits ahead of any bookmarks? I’ve been calling them “anonymous”. What might have been stashes in Git.
comment in response to
post
With time I’ve also reduced my log aliases down to two, one of which is now the default logging template.
comment in response to
post
Otherwise I’ve tried to reduce the aliases down only to those which are run extremely frequently, since otherwise one is likely to forget about them anyway.
comment in response to
post
Ultra offer the FileBrowser app for that. Not amazing but usable.
comment in response to
post
📌
comment in response to
post
Yeah anchoring sadly isn't yet. Firefox is becoming the problematic browser with new standards :(
Does the popover API support hover activation w/o JS?
comment in response to
post
The trick is to stop reading the news and live in blissful ignorance.
comment in response to
post
Done!
comment in response to
post
It’s an unlimited invite tree where poor behaviour by a user reflects up the tree. Similar to torrent trackers.
I’d be happy to invite you, just need an email address. 🙂
comment in response to
post
Could say the same for paths support in the era of native subpath imports and exports.
comment in response to
post
No issues with inwx so far since I switched from Namecheap, and previously Gandi. UI is a bit janky but that seems to be the norm in this space.
comment in response to
post
There are real problems with ActivityPub that others have already talked about a lot, like account portability and how posts propagate across servers. I hope it can continue to improve!
comment in response to
post
There’s a real risk that this platform gets Elon’d in a few years, such is the nature of capitalism, but I don’t care enough about social media to die on this particular hill. Hopefully Atproto’s “credible exit” is, uhm, credible.
comment in response to
post
Bluesky’s identity setup with domains is much better. Very curious where they go with verifications; I’d love something like a web of trust, like LinkedIn vouches but not awful.
comment in response to
post
The status quo means I’m leaving ~30 follows behind, most of whom I could access via a web scraper.
comment in response to
post
The next step for discover feeds may be personalisation via LLMs e.g. I follow hundreds of people but I specify for example “prioritise major product announcements and Dan Abramov”.
comment in response to
post
There’s a lot of dislike for “algorithms”, but that’s actually something pulling me here. I don’t want a chronological feed where every post is weighted equally, it feels like a chore to keep on top of.
comment in response to
post
Bridging should be opt-out for public content. That it’s not is a nonsense that functionally breaks inter-protocol compatibility on the open web.