The Future of Loading on the Web (Chrome Dev Summit 2017)

Ғылым және технология

With the introduction of new performance primitives into the web ecosystem, the future of loading has never looked so bright. In this video, Sam Saccone explores recently shipped performance primitives, cutting edge performance techniques, and a glimpse into what the future of loading on the web may hold.
Check out the rest of the Chrome Dev Summit videos here: goo.gl/ekCoVu
Subscribe to the Google Chrome Developers channel: goo.gl/LLLNvf
event: Chrome Dev Summit 2017; re_ty: Publish;

Пікірлер: 10

  • @gnorberg
    @gnorberg6 жыл бұрын

    Fantastic presentation. Excited for the future of the web!

  • @DenisTRUFFAUT
    @DenisTRUFFAUT6 жыл бұрын

    Great video !

  • @srcspider
    @srcspider6 жыл бұрын

    Props for admitting to the fact that even unminified bundle (with none of the usual special sauce like code splitting) still beats loading granually but you've still omitted a lot of the problems with getting this "magic" to work. For the client to request X, Y, Z from the server the client needs to know the dependency graph (not knowing works, but is slowest since you requerst / parse / excute to find out whats next), and how exactly is the client to know this exactly? If we added a minifest file (ignoring the complexity from having one and the complexity from browsers and servers understanding them (since servers have to bundle everything for you) you have the issue that you're loading one file, then can you start loading. And even with a manifest you're still stuck in the mud when it comes to the "promise" of cherry picking and tree shacking.... that just can't happen with out having dedicated monolithic and inflexible architecture on the server that parses any and all javascript. Let's not even get into "how does the client know what files have actually changed?" with out spamming the server. A bundle is just one check, how much does making 10,000 checks cost? The "problem" you're trying to solve is a real one, however you can just as easily solve it but configuring webpack to split your stuff into more bundles. Code splitting to essentials per page is also already minimal enough that it doesn't really matter IMO (or at least you should in theory be able to get it there in 99% of real world use cases).

  • @filemot25
    @filemot256 жыл бұрын

    Really exciting news

  • @jeffreydwalter
    @jeffreydwalter6 жыл бұрын

    Why the weird naming conventions for the priority stuff? Instead of critical and late, why not highest, high, low, lowest, etc.? Also, why "group" instead of "priority" for the html attribute?

  • @toooes

    @toooes

    6 жыл бұрын

    I wonder too. I think critical and late are explicit terms whereas high/low require context to be meaningful. Edit: From the proposal: We will define a new standard importance attribute that will map to current browser priorities: critical, high, medium, low, unimportant

  • @channel-vt3ib
    @channel-vt3ib6 жыл бұрын

    Hello world,these are my latest development on chrome.

  • @rajashahja8975
    @rajashahja89756 жыл бұрын

    he dont know about web assembly ;-)

  • @DylanIsSoSpooky
    @DylanIsSoSpooky6 жыл бұрын

    Lol I love you google

  • @DylanIsSoSpooky

    @DylanIsSoSpooky

    6 жыл бұрын

    Wait this chrome

Келесі