From: John Thingstad
Subject: Macro dependencies
Date: 
Message-ID: <op.s27l9w06pqzri1@mjolner.upc.no>
This is probably a idea that has cropped up before but here goes.
Why can't the compiler keep track of dependencies between macro's and  
functions.
It seems to me in a ideal world when you change the definition of a macro
all functions that depend on it should be recompiled.
I realize that in systems that have a evaluator this poses a extra  
challenge.
But how about systems like Corman or SBCL that always compile?
They already keep track of the dependencies between functions so it  
doesn't seem
like that far a stretch.
You would also have to keep track of what fasl's changed and store of  
course..
This dependency table can be swapped to disk under normal used and swapped  
in
when needed.

Just my two bits..

-- 
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/
From: John Thingstad
Subject: Re: Macro dependencies
Date: 
Message-ID: <op.s27mpwaapqzri1@mjolner.upc.no>
On Wed, 11 Jan 2006 11:33:22 +0100, John Thingstad  
<··············@chello.no> wrote:

> This is probably a idea that has cropped up before but here goes.
> Why can't the compiler keep track of dependencies between macro's and  
> functions.
> It seems to me in a ideal world when you change the definition of a macro
> all functions that depend on it should be recompiled.
> I realize that in systems that have a evaluator this poses a extra  
> challenge.
> But how about systems like Corman or SBCL that always compile?
> They already keep track of the dependencies between functions so it  
> doesn't seem
> like that far a stretch.
> You would also have to keep track of what fasl's changed and store of  
> course..
> This dependency table can be swapped to disk under normal used and  
> swapped in
> when needed.
>
> Just my two bits..
>

Make that a update-dependencies-macro or something.
Obviously you would want to check that the changes work first!


-- 
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/