Hacker News Reader

Comment by ytygg775 | Hacker News Reader

Clickbaity title. It makes it sound like remote work for everybody everywhere almost broke because of this-or-that thing involving Microsoft and Asus. Not the case. The article is about some obscure issue for a particular company who trusted their IT to be handled by Active Directory in a Microsoft Azure cloud environment, involving Asus home routers. Hardly a general insight.

The root cause for this is the inability of our industry to properly define standards and then enforcing/sticking to them. Everybody just hacks something that kinda-sorta works and whoever has the larger market share is right and everybody else has to suffer, even if they themselves want to do the right thing and do it properly.

Let's face it, we all suck at this and have to pay for it with this kind of meaningless waste of time "troubleshooting".

View on HN

I agree that the title is clickbaity -- I expected something much worse and more intentional than some obscure integration bug between Asus routers' DHCP server and proxy-auto-discovery.

Why our industry typically fails to agree to formal standards is that it takes time and effort to agree upon standards. The W3C is a good example: browser vendors move faster than the W3C could keep up with, so they decided to bypass them.

I don't fully share your negativism towards this: the ability to innovate quickly is important, and even when things get fully standardized, there is no guarantee that every vendor implements them correctly (again, see HTML and just how different things can behave with different browser vendors that all use the same standard).

> the ability to innovate quickly is important

That's a superstition.

No, it's a fact of life. If you don't, then somebody else will and you'll go under.

If you first spend 5 years crafting the perfect invitation to a date for your crush, then meanwhile another guy will have not just asked her out, but by that time they will have rings on their fingers, a house and two kids.

Or you could deem illegal to not follow iso/ietf/w3c unless it is already in a proposal stage.

We may not have the internet we have today but I am not sure it would be for the worse. Some things would have probably slowed down, we may had also avoided the terrible flash/activex/javawebapp period but companies may had pushed for html5 earlier through standardisation process.

No one is going to wait to debate how to do TCP, DNS, or SMTP in an RFC process while not sending bits over the wire.

Most of the core IETF RFCs/BCPs are standardization of things already created. “Prior implementation and testing” is an explicit goal of the IETF process, so I guess you could argue that you’re following that process automatically by creating the initial non-standardized implementation.

That is why I mention that you could deploy it if at least there is a documented proposal.

How would having a published proposal for flash/activeX/Java web app/foobar have avoided those? A company could crank out a draft proposal in a couple of afternoons.

I worked with the author of RFC 1149. He said it didn’t take him long at all to execute it.

HTML5 would not have existed without flash.

Your idea to outlaw imperfect solutions is noble, but misses the fact that mature solutions learn from earlier, imperfect work.

If we’re fining or jailing people for releasing and/or using hacked together code, it either means no progress or rubber-stamp standards bodies that approve everything.

And I feel a little sorry for 13-year-old me, who presumably you would have treated harshly for releasing some BBS software that saw wide adoption when I “should have” invented IPv6 instead.

Trilobites did pretty well. Crocodiles and Greenland sharks are still trucking. Innovation is fun and is strong on the steep bit of the S-curve but it’s not intrinsically superior.

> I don't fully share your negativism towards this: the ability to innovate quickly is important

I fully agree! But that's a different thing. Of course the ability to quickly innovate is important, but you are implying that the current state of affairs is a necessary consequence. It's not.

I also like to "quickly innovate" by hacking some proof of concept code. Exploration is important and fun. I don't write a test first, or no test at all, if I don't even know yet fully what it is that I'm building or how the API needs to look like. But in the long run I can't keep going like that.

These type of bugs / disconnects between systems can have a major productivity impact for teams and organisations. And they are relatively common. :/

I've frequently worked on such issues during my career and the time it takes to identify root cause / effective mitigations can be draining. I've found the key is to focus on creating reproducible test cases and a way to capture environment information in a manner that ensures you are comparing apples to apples. Some systems are so brittle that even adding additional load to capture observability data causes nuances that obscure the root cause. :(

Reply on news.ycombinator.com