Profile avatar
mael.dev
aka arcanis on GitHub. Lead maintainer for @yarnpkg.dev ๐Ÿงถ, staff FE DevX Datadog, ex Facebook, ex Sketchfab. Sharing my life with my wife, two sons, two cats, and three hundred side projects ๐Ÿ“ฆ
156 posts 781 followers 130 following
Getting Started
Active Commenter
comment in response to post
Those layoffs don't make any sense, those companies should be ashamed of themselves. Wishing you all the best.
comment in response to post
Yarn may be a good example for that; we use colors a lot to convey data types: yarnpkg.com/blog#improve...
comment in response to post
You know, what I really don't understand is that you're saying "no, this thing doesn't work, it provides no value" to people who clearly think they already got value out of it. I fully believe *you* didn't see value, but going so far as to invalidate other people's opinion is a little extreme.
comment in response to post
Only a Sith deals in absolutes ๐Ÿคทโ€โ™€๏ธ
comment in response to post
No, the argument rather is that "this thing doesn't have a 100% success rate ergo it's useless" doesn't seem a great argument in a field where pretty much everything is an approximation at various degrees.
comment in response to post
Datadog never sleeps ๐Ÿ™‚ let me know if you're interested in one of our positions careers.datadoghq.com/all-jobs/?ti...
comment in response to post
Whereas we *never* saw Google Search get the search results wrong, lucky for us
comment in response to post
I tried reading the Farseer trilogy again not long ago but grew super frustrated with the main character and his absolute lack of communication skills ๐Ÿ˜ซ
comment in response to post
I'm a fantasy fan but I'm rather going to add the one SF series I love to the list: Gaunt's Ghosts, from Warhammer 40k. Dan Abnett is a master at writing human characters and making us feel their hopes and despairs.
comment in response to post
I felt so betrayed; the Elenium was written years after their conviction, so it seems they never cared to right their wrongs. And they successfully erased this part of their biography until their death. People only found out after stumbling upon old digitalized newspapers.
comment in response to post
I loved D&L Eddings when I was young. A couple of years ago I re-read the book and was surprised to see the heroes doing some casual child abuse in the Elenium series. After digging a bit I discovered D&L had both been convicted of physical child abuse IRL. Tarnished my memories real quick ๐Ÿ˜”
comment in response to post
Sometimes doing less is doing more ๐Ÿ˜„
comment in response to post
I can get behind that
comment in response to post
Oh, the infamous "This is the npm RFC process, not the "all JavaScript package managers" RFC process" RFC ๐Ÿฅด
comment in response to post
Couple of years ago I was thinking about something similar, but there weren't enough use cases at the time github.com/yarnpkg/berr...
comment in response to post
We have a content addressable cache when linking in pnpm mode github.com/yarnpkg/berr...
comment in response to post
That's what we use to support transparent zipped vendors, but the abstraction can be used for a variety of purposes (we also use it to abstract paths across systems).
comment in response to post
Not sure if that can help, but Yarn has a solid virtual filesystem implementation already used in production apps: github.com/yarnpkg/berr...
comment in response to post
Is it a native impl, or a js "mock"?
comment in response to post
No indeed
comment in response to post
Does it have a wasm build? Otherwise afaik stackblitz merely mocks the Node.js internal C bindings with Javascript, so there isn't a vm to run native code.
comment in response to post
I wonder - shouldn't ai tools be integrated into the live ide? You can't prevent their use? Make it part of the evaluation.
comment in response to post
Two kids did that to me! Now I spend this time on Legos, Minecraft, and running around putting out fires ๐Ÿ˜…
comment in response to post
Been there, done that .. the rabbit hole goes deep ๐Ÿ˜„
comment in response to post
I think Python is too mainstream to not get hate, whatever the reason (package management? 2 vs 3? etc). I'd go with Lua - afaik everyone kind agree it's perfect for its niche.
comment in response to post
Exactly that. We used to have a workflow to run CI checks against issues, but having to post statuses as comments was clunky: github.com/arcanis/sher...
comment in response to post
I want status checks on issues, not just PRs ๐Ÿ˜Ÿ
comment in response to post
This is for local development, of course. One published well-written packages should work regardless of the package manager of the consumers (but Corepack doesn't change anything to that).
comment in response to post
What would make a package manager different from any other tool? It's a dependency like any other, it only makes sense to have it locked like any other.
comment in response to post
It's a valid concern - on the other hand, isn't there a chance it could empower more people to build small side projects and experiments?
comment in response to post
It's still unclear to me how a single security release in four years was a burden, or in how that differs from any other Node.js security release. And if it was that difficult, isn't *that* the issue to solve?
comment in response to post
If 30% of Node.js users stated using Corepack in your surveys, EVEN THOUGH it's experimental, and EVEN THOUGH they need to explicitly enable it, and EVEN THOUGH it faced an impressive wave of FUD, it's probably that there's something that isn't achievable just by having a package on npm.
comment in response to post
Still not sure which Corepack users you're talking about that works rejoice about this. It's fairly clear that the value of Corepack is that it's shipped with Node, otherwise as you say there are other options.
comment in response to post
Instead what was shown is pretty much the opposite, up to the point where the TSC allowed borderline harassement strategies to happen without any public repercussions. It's really a shame.
comment in response to post
I'm afraid you're missing the point. Corepack, as a Node.js project, was both a nice UX, and a way to build a bridge between Node.js and communities that felt second-class citizen. Making *this* project succeed was part of the goal, at least for me.
comment in response to post
Will package managers supporting Corepack keep supporting it if they feel second-class citizen? Or would they just accept Node.js will always be toxic for any package-manager-related discussions and move on?
comment in response to post
I tried to write a Typescript-to-cpp converter; it worked okay-ish, but supporting type inference for the various integer sizes was clunky :(
comment in response to post
Awesome you were able to spend resources on this! Did it start as a side experiment? Or did you commit on it from the start?
comment in response to post
Congrats! I've been hoping to see that happen for so long! Can't say I'm not worried about one particular thing though ๐Ÿฅฒ github.com/microsoft/ty...
comment in response to post
The question is whether they'll manage to make our system resilient to malicious agents making their way to a higher position of power. I hope we learn from the US.
comment in response to post
Where is the "Opens an issue without reproduction" arrow? ๐Ÿ˜›
comment in response to post
Il/elle n'a pas l'air trรจs convaincu par les explications ๐Ÿ˜