[racket-dev] racket-5.2.900.1-20120725 read ill-formed code error ?

From: Neil Van Dyke (neil at neilvandyke.org)
Date: Thu Jul 26 17:33:36 EDT 2012

Definitely looks like a Racket bug (or really broken hardware/OS)...

If I start DrRacket 5.2.900.1-20120725 and load up the below file, the 
first few clicks on the Run button give the "ill-formed code error", and 
then, after trying the Run button for 10+ seconds, suddenly Run starts 
working.

#lang racket/base
(require (planet neil/html-template:1:=1))
(html-template (hr (@ (clear "all") (id "foo"))))

If I quit DrRacket and start it back up, the problem is reproducible.

Problem with background optimizing of some kind?

That's all I can spend on this right now; hopefully other people can 
reproduce based on this info.

Neil V.

Neil Van Dyke wrote at 07/26/2012 05:18 PM:
> Anyone know offhand why this error with 5.3 pre-release from yesterday?
>
> UNKNOWN: : read (compiled): ill-formed code [./../src/validate.c:1573]
>   context...:
>    
> /usr/local/racket-5.2.900.1-20120725/lib/racket/collects/racket/private/map.rkt:53:19: 
> loop
>    
> /home/user/.racket/planet/300/5.2.900.1/cache/neil/html-template.plt/1/1/html-template.rkt:103:0: 
> %html-template:transform
>    
> /home/user/.racket/planet/300/5.2.900.1/cache/neil/html-template.plt/1/1/html-template.rkt:523:0 
>
>
> I built Racket from source.  32-bit x86 on Debian Stable.  I don't 
> think there are any old versions of code in "compiled" directories 
> being picked up, but I might have missed some, but I'd expect get an 
> error message about the version mismatch.  PLaneT "html-template:1:1" 
> is currently a bit messy with syntax objects and phases, and might be 
> doing something questionable that 5.2.1 permitted.  This works fine 
> with Racket 5.2.1.
>
> Unfortunately, I don't have time to debug at this moment.  I'm posting 
> now because I understand that a release is close, and I don't know how 
> likely it is that this is a Racket bug.
>
> Neil V.


Posted on the dev mailing list.