bug#12459: Strange issue with circular dependancies

bug#12459: Strange issue with circular dependancies

[rixed]

[-- Attachment #1: Type: text/plain, Size: 2120 bytes --]

While the compiler does manage most circular dependancies, some cases still
cannot be compiled.

For instance, with the attached circ-{a,b,c}.scm the compilation of b fails
with:

% GUILE_LOAD_PATH=. guile-tools compile -Warity-mismatch -Wformat -Wunused-variable -Wduplicate-case-datum -Wbad-case-datum -o circ-b.go circ-b.scm
Backtrace:
In ice-9/boot-9.scm:
2131: 19 [save-module-excursion #<procedure 2648a80 at ice-9/boot-9.scm:2787:17 ()>]
2797: 18 [#<procedure 2648a80 at ice-9/boot-9.scm:2787:17 ()>]
In unknown file:
   ?: 17 [primitive-load-path "circ-a" #f]
In ice-9/eval.scm:
 487: 16 [#<procedure 210ac60 at ice-9/eval.scm:481:4 (exp)> (use-modules # #)]
In ice-9/psyntax.scm:
1091: 15 [expand-top-sequence ((use-modules (circ-b) (circ-c))) () ...]
 976: 14 [scan ((use-modules (circ-b) (circ-c))) () ...]
 270: 13 [scan ((# #) #(syntax-object *unspecified* # #)) () (()) ...]
In ice-9/boot-9.scm:
3258: 12 [process-use-modules (((circ-b)) ((circ-c)))]
 621: 11 [map #<procedure 21d8fe0 at ice-9/boot-9.scm:3258:25 (mif-args)> ((#) (#))]
3259: 10 [#<procedure 21d8fe0 at ice-9/boot-9.scm:3258:25 (mif-args)> ((circ-c))]
2594: 9 [resolve-interface (circ-c) #:select ...]
2519: 8 [#<procedure 21cb760 at ice-9/boot-9.scm:2507:4 (name #:optional autoload version #:key ensure)> # ...]
2786: 7 [try-module-autoload (circ-c) #f]
2131: 6 [save-module-excursion #<procedure 26507b0 at ice-9/boot-9.scm:2787:17 ()>]
2797: 5 [#<procedure 26507b0 at ice-9/boot-9.scm:2787:17 ()>]
In unknown file:
   ?: 4 [primitive-load-path "circ-c" #f]
In ice-9/eval.scm:
 435: 3 [eval #<memoized (define y (@ (circ-b) fb))> ()]
 432: 2 [eval #<memoized (@ (circ-b) fb)> ()]
In unknown file:
   ?: 1 [memoize-variable-access! #<memoized (@ (circ-b) fb)> #f]
In ice-9/boot-9.scm:
 106: 0 [#<procedure 23e9100 at ice-9/boot-9.scm:97:6 (thrown-k . args)> unbound-variable ...]

ice-9/boot-9.scm:106:20: In procedure #<procedure 23e9100 at ice-9/boot-9.scm:97:6 (thrown-k . args)>:
ice-9/boot-9.scm:106:20: In procedure module-lookup: Unbound variable: fb

Note that it compiles if you remove (circ-c) from used-modules in circ-a.scm.


[-- Attachment #2: circ-a.scm --]
[-- Type: text/plain, Size: 134 bytes --]

(define-module (circ-a))
(use-modules (circ-b)
             (circ-c))

(define x 123)
(export x)
(define (fa) (1+ (fb)))
(export fa)


[-- Attachment #3: circ-b.scm --]
[-- Type: text/plain, Size: 72 bytes --]

(define-module (circ-b))

(define (fb) (1+ (@ (circ-a) x)))
(export fb)

[-- Attachment #4: circ-c.scm --]
[-- Type: text/plain, Size: 56 bytes --]

(define-module (circ-c))

(define y (@ (circ-b) fb))
;

bug#12459: Strange issue with circular dependancies

[Ludovic Courtès]

Hi,

rixed@happyleptic.org skribis:

> While the compiler does manage most circular dependancies, some cases still
> cannot be compiled.

It’s really not supposed to work, neither at compile time nor at run time.

Sometimes it turns out to work “by chance” as in this case, or because
there are autoloads to defer the cycle.  But in general, it should
really be avoided.

Won’t fix?  :-)

Ludo’.

bug#12459: Strange issue with circular dependancies

[rixed]

> Won???t fix?  :-)

Fine with me.
I couldn't name a single compiler that can handle circular dependancies.

bug#12459: Strange issue with circular dependancies

[Ludovic Courtès]

tags 12459 wontfix
thanks