WebRTC的问题

Anyone who’s had to move a media catalog from one vendor’s platform to another's knows the frustration and time involved with repeating an entire process of content aggregation all over again. It’s a little like moving all of your stuff from one house to another—some pieces move right over, but sometimes the entire layout is different and things sit in boxes for a long time.

对于我们这些在实时音频/视频领域的人来说, 构建WebRTC解决方案同样令人沮丧, 特别是当您的业务和技术需求增长时. My business has helped more than a few clients move from one WebRTC offering to another, and each process has had different struggles despite the core constructs being fairly identical:

摄像头和麦克风捕捉
智能连接逻辑(P2P、SFU、MCU)
高效的转码处理(VPx到H).264、Opus to AAC等)
归档/回放
穆兴到其他非webbrtc格式(HLS, SRT等)

虽然这个列表并不详尽, 它们可能要求你有任何webbrtc或实时, 低延迟流需求. And for the frustrating part: You won’t likely find all of these requirements fulfilled by any WebRTC server or cloud product, 如果你这么做了, you’ll very likely be using a very specific client-side software development kit (SDK) for each client target (e.g. Web/JavaScript, iOS和Android). 特别是在web/JavaScript方面, 由于Chrome浏览器的不同, 火狐, Safari, 和Edge都实现了WebRTC, these client SDKs are critical to reducing overall development time on a WebRTC project. So, 不管你走哪条路, you’ll have to backtrack and go down another vendor’s WebRTC road whenever you need to move from one WebRTC vendor to another. 你可能会幸运地发现一条通往新路的桥. But one thing is for certain: The client-side SDK for the next WebRTC vendor you choose won’t match the SDK from your prior vendor, 即使在开源产品中也是如此.

现在, 当然, 我需要讲一个闪电侠的轶事, in an effort to find a comparable situation that presented itself in prior periods of streaming history. Flash didn’t have a built-in video player; you couldn’t just pass a video URL to a Flash SWF and have it start playing directly from the Flash Player plugin. Specific code had to be written and compiled into the Flash SWF file to play and manage the video stream. 游戏中没有内置播放/暂停按钮——所有这些都必须自己构建. Enter JWPlayer, FlowPlayer, and any other business that launched from that space. Adobe acknowledged the growing frustration that there was no easy way to add a video player to a Flash project. So, they started the Open Source Media Framework (OSMF) project and made code freely available to anyone wanting to add video into their Flash project. That simplified things, especially for software devs that weren’t video experts as well. 您可以使用OSMF播放来自各种流媒体提供商的视频, and those providers could write custom add-ons or plugins to OSMF to extend its capabilities. Akamai, for example, wrote plug-ins for OSMF to enable features available on their CDN.

那么这和WebRTC有什么关系呢? 正如我之前提到的, each WebRTC vendor will have different client-side SDKs that will require you (or your software development team) to refactor a lot of code to work with the new SDK. Updates across your target destinations (web, Android, iOS) will need to happen. 如果这些更新是相当小的不是很好吗, 而且从百家乐软件管理的角度来看并不昂贵? 就像雷蒙·斯尼奇可能说的那样, I wish this was the part of the story where I was about to now tell you about the Open Source WebRTC Framework, but I’m sad to say that the Baudelaire ophans will have no such gift awaiting them.

My call right now and right here is for the WebRTC titans of industry to formulate an open source WebRTC framework, 图书馆, 随便你怎么称呼它, 并与它们的服务器端对等物一起采用它. It could even be a marketing sales point: “Our product is compatible with the version 1.“WebRTC开放框架的第0位”(当然是我的名字). 现在, there are already community-driven open source code repositories available designed specifically to make WebRTC implementation easier for client-side developers. But they’re all relatively new and have a long way to go to be compatible with different vendor offerings. We need faster progress with viable WebRTC SDKs that don’t tightly tie products to a particular WebRTC server offering.

我将以一种更快乐的方式结束. The good news is that we have finally gotten to a point where desktop and mobile web browsers have aligned in their capabilities to consistently use WebRTC as a core technology for real-time web apps or progressive web apps (PWAs). This means that you won’t need to use a specific browser to enable your tech—or at least you shouldn’t have to if your client-side SDKs support all of them.