[PRL] Joel on Software - Making Wrong Code Look Wrong

[Matthias Felleisen]

On May 11, 2005, at 5:03 PM, Doug Orleans wrote:

> Mitchell Wand writes:
>> And of course the last part, the Rant Against Exceptions is also a 
>> Rant
>> Against Advice.
>
> I don't think this is true (although I wouldn't be surprised if he had
> a different Rant Against Advice).  The gist of his Rant Against
> Exceptions seems to be (translated to Scheme) that the expression
> "(begin (foo) (bar))" may not always result in "bar" being called.
> This is actually a Rant Against Continuations ("foo" might not return
> because it invoked a continuation), and perhaps a Rant Against Macros
> ("begin" might be a macro that ignores its second argument).  But with
> (AspectJ-style) advice, there's no way to prevent "bar" from being
> called (other than, well, throwing an exception in the advice for
> "foo").  You can intercept the call to "bar" and do something before
> or instead of executing the code for "bar", but you can't actually
> prevent the call from happening (in the way that he's ranting about,
> where "foo" never returns).

1. I am told that there are experimental systems (proposals?) where you 
can advice 'begin', 'for', etc.

2. If aspects are weaker than macros, we should build a (parody of the) 
case for macros with the exact arguments that Gregor used for aspects. 
Word for word.

-- Matthias, still reading the dang article