Summary

Specializations using "and" (and/or other macros) don't work properly

Metadata

• Id: 28ba8b51bcf13eda8b45a19694a33bf71777dbf3
• Trac id: 1690
• Type: defect
• Reporter: sjamaan
• Owner:
• Cc:
• Status: closed
• Component: unknown
• Estimated difficulty: medium
• Resolution: fixed
• Priority: major
• Milestone: 5.3
• Version: 5.2.0
• Changetime: 2020-03-27 07:33:12 UTC
• Created: 2020-03-24 19:50:23 UTC
• Keywords: specialization

Description

[[[https://lists.gnu.org/archive/html/chicken-hackers/2020-03/msg00012.html|Reported]|]] by Kon Lovett:

 cat >matches-prob-test.scm <<'HERE'
 (import scheme)
 (import (chicken irregex))
 
 (define (uses-predicate str)
   (and
     (string? str)
     (irregex-match? '(: "foo") str) ) )
 
 (print "uses-predicate: " (uses-predicate ""))
 HERE
 
 csc ./matches-prob-test.scm
 ./matches-prob-test
 #uses-predicate: #f                                                                                             
 
 csc -specialize ./matches-prob-test.scm
 ./matches-prob-test
 #Error: unbound variable: and                                                                                   

Interestingly, this goes all the way back to 4.9.0; in 4.8.0.5 this program (with suitable changes for C4) works as expected with `-specialize`.

It looks like the `and` that's being complained about is the one in the `irregex-match?` specialization in `types.db`. This makes sense if the specialization is the last thing that happens, there's no more macro expansion step and stuff will break.

Changes and comments

[2020-03-24 20:18:11 UTC] sjamaan changed description

[2020-03-24 20:18:11 UTC] sjamaan changed summary

[2020-03-24 20:18:46 UTC] sjamaan changed milestone from 5.4 to 5.3

[2020-03-27 07:33:12 UTC] evhan changed status from new to closed

[2020-03-27 07:33:12 UTC] evhan set resolution to fixed

[2020-03-27 07:33:12 UTC] evhan wrote:

Fixed by d5e8cff27.