Comments on: A ‘When Rails Fails’ Follow-up

By: Onno

Onno — Tue, 17 Nov 2009 16:02:38 +0000

So what's stopping you from writing that debug guide right now?

By: Onno

Onno — Tue, 17 Nov 2009 15:02:38 +0000

So what's stopping you from writing that debug guide right now?

By: T, Rails Programmer

T, Rails Programmer — Mon, 16 Nov 2009 23:57:01 +0000

I agree with the author, that Rails error messages are bad. Usually an error message is just a huge stack trace, with some strange error message at the top. And the place the error originated from isn't even at the top – it's somewhere in the middle of the stack. Now of course I've learned how to read these error messages but at first I was absolutely baffled by them. Some of these errors would be really easy to detect. Like some method expects an integer instead of a string or something (just an example!). The method passes the (wrong) argument to some internal rails methods and it blows up there. Instead there could at least be some sanity checks which spot the error at the place where it happens.
There should be a guide on how to debug Rails errors, among all the other guides which boast how easy and fun development with Rails is. It should be part of the official documentation. Because when a new user read the tutorial and makes a tiny typo error then he is totally screwed and the “fun” of rails is lost for him.

By: Zef Hemel

Zef Hemel — Mon, 16 Nov 2009 18:16:46 +0000

You can extend WebDSL as well, but for many things you'd have to extend the compiler which requires (in our case) knowledge of Stratego/XT. So indeed, there's a trade-off there.

Some things you can do in WebDSL itself, however, you can of course define your own functions and templates. We're also doing a little work on a native class interface, so it's easier to call Java code from WebDSL.

By: Onno

Onno — Sun, 15 Nov 2009 18:26:48 +0000

One issue remains: while a dedicated DSL, such as WebDSL, may be able to solve a lot of the "problems" the author identifies in Rails, it find it hard to believe that WebDSL is just as extensible as Rails. I mean, in a dynamic language I can easily extend the DSL for my own purposes, e.g. by adding a new form helper. I wonder how easy that would be in WebDSL -- and I don't mean that in a sarcastic or rhetorical way, I really do wonder.

Are we going to see some kind of trade off here: flexibility (internal DSL based on a dynamic language) vs. ease of use (external DSL such as WebDSL). Could you comment on that Zef?

By: Ennuyer.net » Blog Archive » Rails Reading - November 15, 2009

Ennuyer.net » Blog Archive » Rails Reading - November 15, 2009 — Sun, 15 Nov 2009 14:00:14 +0000

[...] A ‘When Rails Fails’ Follow-up « I am Zef [...]

By: Elad

Elad — Sat, 14 Nov 2009 20:26:07 +0000

This is a vogue and patronizing argument. "Saying that people need to present a solution before bringing to attention a problem is just silly - a professional debate / opinion article would have included all "failing" frameworks at once, pointing out that this is a cross-platform issue (covering webDSLoser as well). but instead, he chose to practice his opinion by pointing blaming fingers at other implementations (disregarding the fact that it was the most futile attempt of nitpicking i had ever seen) and forgetting to mention that his framework lacks the very basics of which all the other cover.

this is nothing but a cheap, populist and non professional article which its whole purpose was to generate this debate and maybe breath a little life into webDSLoser (srsly dude, restart that one over). This is nothing but a poor attempt for some attention.

By: Elad

Elad — Sat, 14 Nov 2009 20:24:55 +0000

By: Michael

Michael — Sat, 14 Nov 2009 11:44:35 +0000

It doesn't sound like your listening...

Let me repeat the wise words from Xiong Chiamiov:
"Saying that people need to present a solution before bringing to attention a problem is just silly."

By: Michael

Michael — Sat, 14 Nov 2009 11:35:07 +0000

It doesn't sound like your listening...

Let me repeat the wise words from Xiong Chiamiov:
"Saying that people need to present a solution before bringing to attention a problem is just silly."

By: eladmeidar

eladmeidar — Sat, 14 Nov 2009 10:33:24 +0000

what's wrong nancy? did i hurt your feelings?
you are the most hypocrite person i had the chance to read through his stuff, you never seize to amaze me.

By: eladmeidar

eladmeidar — Sat, 14 Nov 2009 10:31:27 +0000

they only need to do so if the ARE WRITING A FRAMEWORK THEMSELVES. blaming others while your very own code does not stand up to your demands is not only silly, but rather hypocrite.

By: Xiong Chiamiov

Xiong Chiamiov — Sat, 14 Nov 2009 01:27:43 +0000

Saying that people need to present a solution before bringing to attention a problem is just silly.

By: Zef Hemel

Zef Hemel — Sat, 14 Nov 2009 01:21:19 +0000

Alright, that's enough. Quit it or I'm blocking you.

By: eladmeidar

eladmeidar — Sat, 14 Nov 2009 01:19:21 +0000

i will believe such integrity exists when i'll see a "a webDSL fails" post, but i won't see one like that. this stupidity has gone far worst than it was ever intended to on my behalf, and i really hope to get to meet Zef one day and show him my appreciation to his professionalism face to face.

By: eladmeidar

eladmeidar — Sat, 14 Nov 2009 01:15:28 +0000

a smart dude said once that "if you live in a house made out of glass, you shouldn't be throwing rocks at other people's houses" - this is exactly the case here, a self-righteous, picky and on top of all misleading post like is exactly what you shouldn't be doing when you own "special" framework sucks ass. @zef should first concentrate his efforts on fixing his own crap before trying to point out stupid and redundant examples for pseudo errors.
My @zef need a bit more assistance when trying to tackle bugs and errors in his own code, but I think that Rails's errors (e.g "Routing Error" in this case) is far more self-explanatory than webDSL complier errors.
This is merely a stupid publicity stunt made only for this cause of pulling some traffic to webDSL, too bad it sucks so bad that I can't believe anyone from the web development industry is going to even try it, never the less go production with it.

By: Xiong Chiamiov

Xiong Chiamiov — Fri, 13 Nov 2009 23:45:16 +0000

He said was that he was examining these other frameworks so that he could improve WebDSL in that area, implying that he is perfectly aware that it needs work.

By: Xiong Chiamiov

Xiong Chiamiov — Fri, 13 Nov 2009 23:43:43 +0000

It's not Ruby that has a problem with error messages, it's Rails. The messages are mostly just default messages from Ruby, which would be helpful if you were writing your application directly in Ruby, rather than have code generated by a framework.

By: lionsburg

lionsburg — Fri, 13 Nov 2009 23:20:42 +0000

Yes, testing could have saved him. However, it is inconsistent behavior from the rest of the framework by not issuing a warning.

For example, ActiveRecord when presented with this:
User.find(:all, :some_unknown_arg => 1)

Will throw this exception:

ArgumentError: Unknown key(s): some_unknown_arg

Why is it not reasonable to expect link_to to behave in similar fashion?

By: lionsburg

lionsburg — Fri, 13 Nov 2009 23:17:10 +0000

What you are commenting on is completely reasonable and there's no reasons why the framework couldn't have that checking it in.

Another response to your Category #5 "Where are your tests?" is: Tests doesn't automagically point to where a problem is. It may get you close. Most of the examples you provide the framework could pretty easily issue warnings to help you zero directly in on the typos.

Good luck with the crazies and religious zealots.