[bug #30169] Auto-compilation failure of a module doesn't lead to a failure [1.9.11]

[bug #30169] Auto-compilation failure of a module doesn't lead to a failure [1.9.11]

[Ludovic Courtès]

URL:
  <http://savannah.gnu.org/bugs/?30169>

                 Summary: Auto-compilation failure of a module doesn't lead
to a failure [1.9.11]
                 Project: Guile
            Submitted by: civodul
            Submitted on: Thu 17 Jun 2010 01:06:37 PM GMT
                Category: None
                Severity: 3 - Normal
              Item Group: None
                  Status: None
                 Privacy: Public
             Assigned to: None
             Open/Closed: Open
         Discussion Lock: Any

    _______________________________________________________

Details:

Consider this example, where auto-compilation of (foo) fails:

#v+
$ cat foo.scm
(define-module (foo))
(open

$ cat bar.scm
(use-modules (foo))
(display "hello!\n")

$ guile -L . bar.scm
;;; note: source file bar.scm
;;;       newer than compiled
/home/ludo/.cache/guile/ccache/2.0-0.R-LE-8/home/ludo/src/guile/bar.scm.go
;;; note: autocompilation is enabled, set GUILE_AUTO_COMPILE=0
;;;       or pass the --no-autocompile argument to disable.
;;; compiling bar.scm
;;; compiling ./foo.scm
;;; WARNING: compilation of ./foo.scm failed:
;;; key read-error, throw args ("scm_i_lreadparen" "./foo.scm:3:1: end of
file" () #f)
;;; WARNING: compilation of bar.scm failed:
;;; key read-error, throw_args ("scm_i_lreadparen" "./foo.scm:3:1: end of
file" () #f)
hello!

$ echo $?
0
#v-

Guile should instead abort as soon as auto-compilation fails.

Thanks,
Ludo'.




    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?30169>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/

Re: [bug #30169] Auto-compilation failure of a module doesn't lead to a failure [1.9.11]

[Andy Wingo]

Hello,

On Thu 17 Jun 2010 15:06, Ludovic Courtès <INVALID.NOREPLY@gnu.org> writes:

> Guile should instead abort as soon as auto-compilation fails.

Guile can fail autocompilation due to eval-when issues for syntax
helpers. Do you want to make Guile fail in this case?

Andy
-- 
http://wingolog.org/

Re: [bug #30169] Auto-compilation failure of a module doesn't lead to a failure [1.9.11]

[Ludovic Courtès]

Hi,

Andy Wingo <wingo@pobox.com> writes:

> On Thu 17 Jun 2010 15:06, Ludovic Courtès <INVALID.NOREPLY@gnu.org> writes:
>
>> Guile should instead abort as soon as auto-compilation fails.
>
> Guile can fail autocompilation due to eval-when issues for syntax
> helpers. Do you want to make Guile fail in this case?

Sorry, I don’t understand.  I want Guile to fail in the example I gave.

Ludo’.

Re: [bug #30169] Auto-compilation failure of a module doesn't lead to a failure [1.9.11]

[Andy Wingo]

On Fri 18 Jun 2010 16:52, ludo@gnu.org (Ludovic Courtès) writes:

> Andy Wingo <wingo@pobox.com> writes:
>
>> On Thu 17 Jun 2010 15:06, Ludovic Courtès <INVALID.NOREPLY@gnu.org> writes:
>>
>>> Guile should instead abort as soon as auto-compilation fails.
>>
>> Guile can fail autocompilation due to eval-when issues for syntax
>> helpers. Do you want to make Guile fail in this case?
>
> Sorry, I don’t understand.  I want Guile to fail in the example I gave.

Some files may be interpreted, but fail to compile. They will cause
autocompilation to fail. To be specific, from NEWS:

    ** Functions needed by macros at expand-time need to be present at
       expand-time.

    For example, this code will work at the REPL:

      (define (double-helper x) (* x x))
      (define-macro (double-literal x) (double-helper x))
      (double-literal 2) => 4

    But it will not work when a file is compiled, because the definition of
    `double-helper' is not present at expand-time. The solution is to wrap
    the definition of `double-helper' in `eval-when':

      (eval-when (load compile eval)
        (define (double-helper x) (* x x)))
      (define-macro (double-literal x) (double-helper x))
      (double-literal 2) => 4

    See the documentation for eval-when for more information.

So, you want Guile to fail faster for this error; how do you distinguish
it from eval-when errors as above? Or would you prefer that eval-when
errors lead to failures as well?

Andy
-- 
http://wingolog.org/

[bug #30169] Auto-compilation failure of a module doesn't lead to a failure [1.9.11]

[Ludovic Courtès]

Follow-up Comment #1, bug #30169 (project guile):

(Follow-up from bug-guile.)

Andy says:

> Some files may be interpreted, but fail to compile.

OK.

[...]

> So, you want Guile to fail faster for this error; how do you distinguish
> it from eval-when errors as above? Or would you prefer that eval-when
> errors lead to failures as well?

Do you mean that, for programs written to be interpreted, which lack proper
‘eval-when’ clauses, the idea was to fall back to interpretation when
auto-compilation fails?  That would indeed be friendlier to programs written
for previous versions of Guile.

However, in the example I gave, somehow, an error should occur.  With 1.8 you
would get an error when reading ‘foo.scm’, whereas here it’s just
silently ignored, which sounds wrong.

Ludo'.

    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?30169>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/

Re: [bug #30169] Auto-compilation failure of a module doesn't lead to a failure [1.9.11]

[Andy Wingo]

Hi,

On Sun 20 Jun 2010 16:23, Ludovic Courtès <INVALID.NOREPLY@gnu.org> writes:

>> So, you want Guile to fail faster for this error; how do you distinguish
>> it from eval-when errors as above? Or would you prefer that eval-when
>> errors lead to failures as well?
>
> Do you mean that, for programs written to be interpreted, which lack proper
> ‘eval-when’ clauses, the idea was to fall back to interpretation when
> auto-compilation fails?  That would indeed be friendlier to programs written
> for previous versions of Guile.

Yes, that was the goal.

> However, in the example I gave, somehow, an error should occur.  With 1.8 you
> would get an error when reading ‘foo.scm’, whereas here it’s just
> silently ignored, which sounds wrong.

Indeed.

Andy
-- 
http://wingolog.org/

[bug #30169] Auto-compilation failure of a module doesn't lead to a failure [1.9.11]

[Andy Wingo]

Follow-up Comment #2, bug #30169 (project guile):

You know, we might consider this to be the same problem as we see here:

$ guile -L .
scheme@(guile-user)> (use-modules (foo))
;;; note: autocompilation is enabled, set GUILE_AUTO_COMPILE=0
;;;       or pass the --no-autocompile argument to disable.
;;; compiling /tmp/foo.scm
;;; WARNING: compilation of /tmp/foo.scm failed:
;;; key read-error, throw args ("scm_i_lreadparen" "/tmp/foo.scm:3:1: end of
file" () #f)
Throw to key `read-error':
ERROR: In procedure scm_i_lreadparen:
ERROR: /tmp/foo.scm:3:1: end of file
Entering the debugger. Type `bt' for a backtrace or `c' to continue.
0 debug> c
ERROR: In procedure scm_i_lreadparen:
ERROR: /tmp/foo.scm:3:1: end of file
scheme@(guile-user)> (use-modules (foo))
scheme@(guile-user)> 

The second (use-modules (foo)) should not succeed.

    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?30169>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/