Hacker News Reader

Comment by Joe8Bit | Hacker News Reader

I understand there a lots of governance/no technical differences, but is there a clear answer as to why Joyent can't use 'fast forward' Node to the current IO.js stable? Are there major discrepancies between the roads the two teams have taken that would necessitate a 'merge' of the two projects?

View on HN

Yes. And this is the reason I can't take people who advocate io.js seriously. They either didn't do their homework, or have a bizarre set of requirements from software. Joyent did lots of improvements in node.js observability with DTrace and mdb providers. This was one of the very first thing that broke after io.js forked.

Yes, releasing a new release every week is more important than being to debug, profile, and understand your application.

As far as I understood, removing MDB was an unfortunate side-effect of upgrading V8 to a newer version[0], and while I can see this would be an issue for some people, I would argue that bumping V8 and the performance and core language improvements that come with it is worth it! But I perfectly see you're point.

Also, this is going to be an issue for Node core too eventually, unless they plan to stay on an old version of V8 indefinitely.

Not sure what you're referring to about DTrace issues in IO.JS, could you provide some guidance?

0: https://github.com/iojs/io.js/pull/1023

DTrace works exactly like mdb, it requires knowledge about V8 internals.

The problem is haphazardly upgrading the V8 version without thinking about the consequences. This is not how you do release engineering, especially since io.js is advertised as stable.

Stable doesn't mean "it works for me", it means you provide some formal guarantees to downstream consumers about the stability of exposed interfaces.

I'm very well aware that some (most?) people don't care about things like good introspection support and stability, but it's pretty pathetic to see factual information being downvoted because people don't agree with the message.

I think "haphazardly" might be a little harsh, my understanding is that the IO.js core team fully understood the introspection regressions that would occur when bumping V8; but they took the descison that improved runtime performance combined with native support MANY new language features meant the regressions were acceptable. I agree fwiw.

They favoured the many (who would use the new V8's features) over the few (who valued DTrace etc)

I haven't heard any real claims as to IO.js instability, and have been using it production at scale for several months myself.

I'd rather have all of the bug-fixes and improvements from the new v8 engine.

I'm able to debug, profile and understand my application just fine under io.js. You don't need DTrace or MDB to do this.

Reply on news.ycombinator.com