Node devs leave bug in url.parse() citing browser history | Hacker News Reader

Node devs leave bug in url.parse() citing browser history

github.com

46 pointswlll19 comments

I don't see any problems with this. The devs also have to balance the weight of breaking current applications using their existing url parsing. It also doesn't seem an unreasonable argument that users should expect node to carry on existing conventions from js in browsers. I'd make the same choice.

I wouldn't make the same choice. Node is young enough that a minor change like this with proper warning to developers should be expected. If Node's version was >1.0.0 I'd expect this bug to have been fixed years ago. It's completely subjective, but it gives further credence to opinions that Node is a toy. Not that I agree with it.

I see that you also hate freedom.

We should hang out.

Heh, I read a blog post a few days ago about the differences between various URL parsing/slicing implementations.

http://tantek.com/2011/238/b1/many-ways-slice-url-name-piece...

(Quick reference image: http://farm7.static.flickr.com/6203/6082913622_c953b1fc96_o.... )

Clearly mountains out of mole hills

Best Comment: @chjj "Is it me, or is this argument over a single colon?"

https://github.com/joyent/node/pull/1580#issuecomment-192688...

The proposed addition of an additonal 'scheme' return value, that dosen't include the colon seems reasonable enough. Yes its kind of unnecessary but a more respectful conclusion then, inciting this argument and tarnishing the node community any more than it already is.

What's this, HN is now personal army for pissing matches?

Nothing to see here, move along.

It's a discussion about a programming issue that could be of interest to hackers everywhere who run or participate in OS projects with implications on how to respond to this sort of situation (expected behaviour vs. standards). Seems to me like there's something to see, and that's why I submitted it.

I'd leave the "bug" in too... much in the same way every web-protocol software since forever has kept the misspelling of 'referer' intact.

Breaking the commonly used standard isn't a good idea. As long as the behavior is documented and consistent, don't modify it for pedantic reasons, particularly when the modification will break existing code and cause confusion for people who expect the other behavior due to it working that way in browsers.

The difference I guess is that rather than "referer" having only one setup (the way it's specced and the way it's implemented are identical), in this case there are two competing methods: The way browsers do it, and the way every other language library does it (which is, incidentally, the way the spec says it should be done).

Yeah my analogy certainly isn't a perfect one and that is a good point, but for node.js I'd still argue that being compatible with JavaScript running on browsers (even if that means being 'wrong') is much more important than being compatible with other languages.

It's a reasonable choice to make even if I would have gone the other way.

If you don't plan on supporting this library long term or if you think that the node ecosystem isn't going to be long lived then why cause pain now in order to be compliant with the RFC? Right now it can match the browser bug and not confuse web designers moving to doing server side code thanks to Node.

It is just a single colon, but in my experience this kind of debate comes up early in an ecosystems development and the decisions, especially ones that get a lot of attention, set the culture's direction pretty quick.

I don't think it's a good idea long term but it's not my library so it's not my call, and I can see the pragmatism behind the authors choice.

As a node user I don't see this as a bug. It's a bit pissy to submit it with this title. I think being inline with browsers and expected node behaviour is more important than with python/ruby/java/perl/the rfc, which is about the only argument Jordan came up with.

Jordan said in his last reply: >I would recommend folks stop updating this issue since you'll likely just get stuck in a loop of providing data and having the data be discarded

This is far too quick to jump to accusing isaac of ignoring answers, reading the thread he and other submitters say they don't believe other languages doing it a certain way is reasons to change the behaviour. So Jordan accuses him of ignoring his argument?

This is hilarious.

I did this because I hate freedom.

What has Freedom got to do with this? Not arguing with you, I just don't understand where your comment is coming from.

It's a sarcastic joke. Check out this site maintained by the commenter http://blog.izs.me/

Well, because we're apparently contribution-hostile. https://twitter.com/ericflo/status/107963620159266816

(I'm isaacs, node core dev and author of the url parsing module, and closer of the bug in the OP.)

Not "apparently". A guy showed up wanting to contribute, and based on the response he got, he perceived a "foot-stomp and 'you shall not pass'."

It doesn't matter whether you think you're contribution-hostile, it matters whether the guy who tried to contribute thought you came off as hostile, which he clearly did.

It even started off poorly.

Getting a 'why?' and an immediate close certainly would give me a poor first impression, as a developer sending a patch for something I thought was an actual bug.

Reply on news.ycombinator.com