From: cjon@... (Cullen J O'neill)
Date: 2000-12-02T06:40:06+09:00
Subject: [ruby-talk:6718] Re: rescue clause affecting IO loop behavior

Conrad Schneiker (schneik@us.ibm.com) wrote:
> I think you probably want to get such warnings only when explicitly 
> requested (-w) so that applications don't start issuing these things to 
> unsuspecting users/customers following Ruby upgrades. It would also be 
> useful if such a mechanism worked with a "code check only" mode, so that 
> information-gathering runs against the RAA and so on could be made, so as 
> to estimate the impact of widely desired changes that are not completely 
> backwards compatible.

I think it could start out that way, but (maybe when changing minor
numbers [i.e.: from 1.a.x to 1.b.x]) eventually it has to be on by default
so that we can move toward eliminating that disparaged feature from the
language.  I'm speaking here independent of whether the below features
are problems or not.  We need some way to continue changing the language
as it becomes popular with minimum pain all around.

> Meanwhile, you could read "else" as "[you should] otherwise do" to make it 
> read somewhat more naturally. 
>
> # I think using else with unless should also be disparaged.
>
> I like unless better than if (! ...), and I'd want to preserve the option 
> of adding else (something ...) for [hypothetically speaking] quickly 
> patching the production code of others, when you want to minimize chances 
> for errors in rewriting or want to minimize diff output.  JMHO.

Ok, this is the closest to a good argument I've seen for this feature.  
Still, do we want to encourage leaving in stinky code (stinky in the
code refactoring sense) in this manner?  There is a middle path between
a fascist langauge and Perl's excessive (IMHO) TMTOWTDIness you know... ;)


Cullen J O'Neill
--
cjon@engin.umich.edu