tbroyer

tbroyer, 17 days ago

@Larvitz @nixCraft Wasn't that exactly the point? That CentOS moved from being a downstream of RHEL to an upstream? (to somehow crowdsource testing 😉)

tbroyer, 14 days ago

@josh Well, it's not just react vs web components.

They also rearchitected their code; and as a side effect of moving away from react, they can have code optimized for the most recent Edge when react needs to support many more browser engines and versions (something you definitely want for most web apps).

tbroyer, 15 days ago

@thudfactor @davatron5000 Shameless plug: https://blog.ltgt.net/web-component-libs-benefits/

tbroyer, 14 days ago

@mayank Well, technically, dashed names don't create "custom elements" either, but the same kind of "unknown elements".

BTW, I just noticed this in the spec: https://html.spec.whatwg.org/multipage/custom-elements.html#custom-elements-upgrades-examples
> Because element definition can occur at any time, a non-custom element could be created, and then later become a custom element after an appropriate definition is registered. We call this process "upgrading" the element, from a normal element into a custom element.

So a dashed name, per spec, creates a "non-custom element" (aka unknown element), and a "custom element" is one that has been registered.

tbroyer, 14 days ago

@mayank Oh sure, but due to https://html.spec.whatwg.org/multipage/syntax.html#syntax-tag-name so would any element with a : or _ in its name (and those are just examples, HTML is very lenient when it comes to parsing tag names: https://html.spec.whatwg.org/multipage/parsing.html#tag-name-state), but those can't be registered in the "custom element registry" so they can't be turned in "defined custom elements" (which everyone just calls "custom elements" I believe)

tbroyer, 14 days ago

@mayank I have another one for you: are "non-defined custom elements" really custom elements? Or more "unknown elements that happen to have a hyphen in their name"? (because if you only use them to match them in a CSS selector then you could use any name, even without an hyphen)

tbroyer, 13 days ago

@bagder The problem is not the tools themselves (not entirely at least because they have many shortcomings, like not accounting for Debian/Ubuntu fixes backports) but how people (don't) analyze the results.
We do use vulnerability scanners (sending SBOMs to Dependency-Track) but this clearly requires work to analyze the results and determine if you're actually vulnerable (false positive, non-exploitable vuln, only if configured in some specific way, etc.)
Contacting project maintainers for assistance without even some prior analysis is just plain wrong I 💯 agree!

tbroyer, 1 month ago

@ayo HOCON is quite good as well
https://github.com/lightbend/config/blob/main/HOCON.md

tbroyer, 3 months ago

@zachleat @develwithoutacause Probably because no built-in element would ever throw when changing an attribute.
State is derived from attributes, but attributes don't directly "set" state, and it's not abnormal to sometimes be in some sort of "invalid state" and you have to cope for it.

Would you like it if you had to be careful updating min, max and value in proper order or it would throw?

See https://blog.ltgt.net/web-component-properties/ where I touch the subject a bit (among other things)

tbroyer, 18 days ago

@nixCraft Particularly as many fedi clients do support it!

tbroyer, 12 days ago

@Lukew @westbrook Ah yes, I do remember that in 2021 the french website https://vitemadose.covidtracker.fr (to easily book covid vaccination appointments) had (and probably still has) a similar issue: the CSS ended up duplicated between the main page and the shadow dom. This was due to import styles from "style.scss" both injecting the styles in the HTML and returning them as string (so they could be used as unsafeCss with Lit).
I did report this to Vite at the time and indeed the answer was to use ?inline (might have been added specifically in response to that issue): https://github.com/vitejs/vite/issues/3246
That was 3 years ago though, I'd have hoped it would have changed…

tbroyer, 13 days ago

@westbrook I don't use vite but know teams at work who do so I'm curious. Could you name a few of those things it does?

tbroyer, 20 days ago

Quelle version de Compose ? Parce que "it works on my machine", que le conteneur tourne toujours (detached) ou non.

$ docker compose run keycloak true  
$ docker compose run -d keycloak cat  
$ docker compose ps --all  
<montre les 2 conteneurs, un Up, l'autre Exited>  
$ docker compose down --remove-orphans  
<arrête le conteneur qui tourne, supprime les 2 conteneurs>  
$ docker compose version  
Docker Compose version 2.27.0  

tbroyer, 20 days ago

@aeris docker compose down --remove-orphans?
https://docs.docker.com/reference/cli/docker/compose/down/

tbroyer, 13 days ago

@fmuc @nixCraft Nah, that's for frontend web developers fan of web components: https://developer.mozilla.org/en-US/docs/Web/API/Web_components/Using_shadow_DOM

tbroyer, 9 days ago

@tanepiper AFAIK the most critical level is not optional, you can only disable lower levels (and given how irresponsibly the Paris prefecture abused it to advertize for the olympic games' opening ceremony pass, I bet may parisians disabled those lower levels already).

tbroyer, 15 days ago

@niclake @eleventy Well, GitHub Pages?

https://github.com/tbroyer/blog.ltgt.net/blob/main/.github/workflows/github-pages.yml

tbroyer, 4 months ago

@dutchcelt Do you think Apple continues to invest in Safari because of some sunk-costs fallacy? They could have already switched to Chromium a long time ago, so I don't think this is going to happen.

As for whether Safari will lose its dominance on iOS, I'm not sure either: people don't care about the browser engine, they care about integrations and they (already) use Chrome on iOS because of how it can sync their data with their Google account, whether WebKit or Chromium.

tbroyer, 5 days ago

@zachleat Did use one to have more grip, but it was a bit too big. Removed it a few weeks ago and I don't think I'll get back (well, except it allows me to also attach an accessory… which I only use once in a while… (Xiaomi photography kit))

tbroyer, 12 days ago

@tomayac @patrickbrosset We've been using subgrid at work in an internal knowledgebase site since before it came to Chrome. It goes like this:

<body>  
 <header>  
 <div>site title  
 <div>theme toggle  
 <div>  
 <div>search  
 <div>site version  
 <nav>main menu  
 <main>  
 <h1>page title  
 <div>main content  
 <div>author, date, tags, outline  
 <footer>  
 <div>left  
 <div>right  

and we want to render in 1, 2 or 3 columns. We could remove the <header>, <div>, <main> and <footer> grouping level and just use grid, but using subgrid seems like a better fit here.

tbroyer, 20 days ago

@melix …and every single place you need to update (particularly unit tests that don't use DI) whenever dependencies change: code just fails to compile.