[racket-dev] massive penalty for (except-in (planet ...) ...) require ?

From: Carl Eastlund (cce at ccs.neu.edu)
Date: Thu Feb 17 17:22:57 EST 2011

Have you tried this with non-planet files?  I'm curious whether the
planet aspect is really a factor here, or just except-in.

Carl Eastlund

On Thu, Feb 17, 2011 at 5:18 PM, John Clements
<clements at brinckerhoff.org> wrote:
> On Feb 17, 2011, at 2:12 PM, Carl Eastlund wrote:
>> I don't know precisely what is going on here, but I know from looking
>> at expanded code that straightforward module requires become
>> straightforward #%requires, while modifications like renaming wind up
>> expanding out each individual binding as a separate #%require.  So it
>> seems plausible that except-in, rename-in, etc. might seriously impact
>> compile time.
> Also, FWIW: this is a hard-link planet package, not a normally installed one.
> John
>> On Thu, Feb 17, 2011 at 5:04 PM, John Clements
>> <clements at brinckerhoff.org> wrote:
>>> Running a particular file has been crushingly slow for me, and I finally traced it down
>>> (apparently) to the addition of an "except-in" wrapper around a planet require.  Adding the
>>> wrapper changes a consistently sub-1-second compile-and run into a big thrashing
>>> memory-fest. Is there some kind of invasive renaming that must occur when an "except-in"
>>> annotation is added to a require?
>>> John

Posted on the dev mailing list.