guix-pull prints stack trace

guix-pull prints stack trace

[Alex Vong]

Hello,

It seems the commit c41d97b causes guix-pull to print stack trace.

Here is the stack trace:
===========================================================================
In ice-9/psyntax.scm:
 989: 19 [scan ((define-module (gnu packages tex) #:use-module ...)) () ...]
 279: 18 [scan ((#(syntax-object let # ...) (#) (# #) ...)) () ...]
In ice-9/eval.scm:
 411: 17 [eval # ()]
In ice-9/boot-9.scm:
2951: 16 [define-module* (gnu packages tex) #:filename ...]
2926: 15 [resolve-imports ((# # license:) (#) (#) (#) ...)]
2864: 14 [resolve-interface (gnu packages pdf) #:select ...]
2789: 13 [#<procedure 84d300 at ice-9/boot-9.scm:2777:4 (name #:optional autoload version #:key ensure)> # ...]
3065: 12 [try-module-autoload (gnu packages pdf) #f]
2401: 11 [save-module-excursion #<procedure 5cedb40 at ice-9/boot-9.scm:3066:17 ()>]
3085: 10 [#<procedure 5cedb40 at ice-9/boot-9.scm:3066:17 ()>]
In unknown file:
   ?: 9 [primitive-load-path "gnu/packages/pdf" ...]
In ice-9/eval.scm:
 505: 8 [#<procedure 79a6a0 at ice-9/eval.scm:499:4 (exp)> (define-module # # ...)]
In ice-9/psyntax.scm:
1106: 7 [expand-top-sequence ((define-module # # # ...)) () ((top)) ...]
 989: 6 [scan ((define-module (gnu packages pdf) #:use-module ...)) () ...]
 279: 5 [scan ((#(syntax-object let # ...) (#) (# #) ...)) () ...]
In ice-9/eval.scm:
 411: 4 [eval # ()]
In ice-9/boot-9.scm:
2951: 3 [define-module* (gnu packages pdf) #:filename ...]
2926: 2 [resolve-imports ((# # license:) (#) (#) (#) ...)]
2867: 1 [resolve-interface (gnu packages javascript) #:select ...]
In unknown file:
   ?: 0 [scm-error misc-error #f ...]

ERROR: In procedure scm-error:
ERROR: no code for module (gnu packages javascript)
builder for `/gnu/store/58ji3i7a0dl2sgf39m16l90alnyyvc9j-guix-latest.drv' failed with exit code 1
guix pull: error: build failed: build of `/gnu/store/58ji3i7a0dl2sgf39m16l90alnyyvc9j-guix-latest.drv' failed
===========================================================================

Looking at the patch, I think the problem is in
`#:use-module (gnu packages javascript)'.

Cheers,
Alex

Re: guix-pull prints stack trace

[Taylan Ulrich Bayırlı/Kammer]

Alex Vong <alexvong1995@gmail.com> writes:

> Hello,
>
> It seems the commit c41d97b causes guix-pull to print stack trace.
>
> Here is the stack trace:
> ===========================================================================
> In ice-9/psyntax.scm:
>  989: 19 [scan ((define-module (gnu packages tex) #:use-module ...)) () ...]
>  279: 18 [scan ((#(syntax-object let # ...) (#) (# #) ...)) () ...]
> In ice-9/eval.scm:
>  411: 17 [eval # ()]
> In ice-9/boot-9.scm:
> 2951: 16 [define-module* (gnu packages tex) #:filename ...]
> 2926: 15 [resolve-imports ((# # license:) (#) (#) (#) ...)]
> 2864: 14 [resolve-interface (gnu packages pdf) #:select ...]
> 2789: 13 [#<procedure 84d300 at ice-9/boot-9.scm:2777:4 (name #:optional autoload version #:key ensure)> # ...]
> 3065: 12 [try-module-autoload (gnu packages pdf) #f]
> 2401: 11 [save-module-excursion #<procedure 5cedb40 at ice-9/boot-9.scm:3066:17 ()>]
> 3085: 10 [#<procedure 5cedb40 at ice-9/boot-9.scm:3066:17 ()>]
> In unknown file:
>    ?: 9 [primitive-load-path "gnu/packages/pdf" ...]
> In ice-9/eval.scm:
>  505: 8 [#<procedure 79a6a0 at ice-9/eval.scm:499:4 (exp)> (define-module # # ...)]
> In ice-9/psyntax.scm:
> 1106: 7 [expand-top-sequence ((define-module # # # ...)) () ((top)) ...]
>  989: 6 [scan ((define-module (gnu packages pdf) #:use-module ...)) () ...]
>  279: 5 [scan ((#(syntax-object let # ...) (#) (# #) ...)) () ...]
> In ice-9/eval.scm:
>  411: 4 [eval # ()]
> In ice-9/boot-9.scm:
> 2951: 3 [define-module* (gnu packages pdf) #:filename ...]
> 2926: 2 [resolve-imports ((# # license:) (#) (#) (#) ...)]
> 2867: 1 [resolve-interface (gnu packages javascript) #:select ...]
> In unknown file:
>    ?: 0 [scm-error misc-error #f ...]
>
> ERROR: In procedure scm-error:
> ERROR: no code for module (gnu packages javascript)
> builder for `/gnu/store/58ji3i7a0dl2sgf39m16l90alnyyvc9j-guix-latest.drv' failed with exit code 1
> guix pull: error: build failed: build of `/gnu/store/58ji3i7a0dl2sgf39m16l90alnyyvc9j-guix-latest.drv' failed
> ===========================================================================
>
> Looking at the patch, I think the problem is in
> `#:use-module (gnu packages javascript)'.

The bug I introduced to guix pull (while speeding it up) that has a tiny
chance of causing a thread race related error has never been fixed.
Maybe it's related to that.

As of e714797 I can't reproduce the error; guix pull runs fine to
completion.  Possibly a commit after c41d97b "fixed" the issue again by
changing the order of autoloads or some such.  But it could also be
related to the number of cores on my machine, how much you moved around
your pointer device recently, or the phase of the moon. :-\

If you still hit the error, or anyone else does, I'll get to work on it
ASAP.

Otherwise, I'm quite out of the loop but IIRC there was some talk of an
entirely different guix pull mechanism, that uses git pull or so?  If
so and if this bug continues to trigger extremely rarely, then I'm not
sure if it would be worth the effort to make bigger changes to the
current 'guix pull' mechanism.

Taylan


P.S.: Feel free to contact me on Freenode (nick 'taylan') if you want to
bypass email for quicker communication.

Re: guix-pull prints stack trace

[Alex Vong]

taylanbayirli@gmail.com (Taylan Ulrich "Bayırlı/Kammer") writes:

> Alex Vong <alexvong1995@gmail.com> writes:
>
[...]
>
> The bug I introduced to guix pull (while speeding it up) that has a tiny
> chance of causing a thread race related error has never been fixed.
> Maybe it's related to that.
>
> As of e714797 I can't reproduce the error; guix pull runs fine to
> completion.  Possibly a commit after c41d97b "fixed" the issue again by
> changing the order of autoloads or some such.  But it could also be
> related to the number of cores on my machine, how much you moved around
> your pointer device recently, or the phase of the moon. :-\
>
> If you still hit the error, or anyone else does, I'll get to work on it
> ASAP.
>
Yes, commit e714797 'gnu: pdf: Remove erroneous module import.' seems to
have fixed the issue again. Thanks for your attention. I don't know
there is a potential race condition in guix-pull. I will keep this in
mind.

> Otherwise, I'm quite out of the loop but IIRC there was some talk of an
> entirely different guix pull mechanism, that uses git pull or so?  If
> so and if this bug continues to trigger extremely rarely, then I'm not
> sure if it would be worth the effort to make bigger changes to the
> current 'guix pull' mechanism.
>
> Taylan
>
>
> P.S.: Feel free to contact me on Freenode (nick 'taylan') if you want to
> bypass email for quicker communication.

Thanks,
Alex

Re: guix-pull prints stack trace

[Leo Famulari]

On Sat, Sep 10, 2016 at 01:14:10PM +0800, Alex Vong wrote:
> Hello,
> 
> It seems the commit c41d97b causes guix-pull to print stack trace.
> 
> Here is the stack trace:
> ===========================================================================
> In ice-9/psyntax.scm:
>  989: 19 [scan ((define-module (gnu packages tex) #:use-module ...)) () ...]
>  279: 18 [scan ((#(syntax-object let # ...) (#) (# #) ...)) () ...]
> In ice-9/eval.scm:
>  411: 17 [eval # ()]
> In ice-9/boot-9.scm:
> 2951: 16 [define-module* (gnu packages tex) #:filename ...]
> 2926: 15 [resolve-imports ((# # license:) (#) (#) (#) ...)]
> 2864: 14 [resolve-interface (gnu packages pdf) #:select ...]
> 2789: 13 [#<procedure 84d300 at ice-9/boot-9.scm:2777:4 (name #:optional autoload version #:key ensure)> # ...]
> 3065: 12 [try-module-autoload (gnu packages pdf) #f]
> 2401: 11 [save-module-excursion #<procedure 5cedb40 at ice-9/boot-9.scm:3066:17 ()>]
> 3085: 10 [#<procedure 5cedb40 at ice-9/boot-9.scm:3066:17 ()>]
> In unknown file:
>    ?: 9 [primitive-load-path "gnu/packages/pdf" ...]
> In ice-9/eval.scm:
>  505: 8 [#<procedure 79a6a0 at ice-9/eval.scm:499:4 (exp)> (define-module # # ...)]
> In ice-9/psyntax.scm:
> 1106: 7 [expand-top-sequence ((define-module # # # ...)) () ((top)) ...]
>  989: 6 [scan ((define-module (gnu packages pdf) #:use-module ...)) () ...]
>  279: 5 [scan ((#(syntax-object let # ...) (#) (# #) ...)) () ...]
> In ice-9/eval.scm:
>  411: 4 [eval # ()]
> In ice-9/boot-9.scm:
> 2951: 3 [define-module* (gnu packages pdf) #:filename ...]
> 2926: 2 [resolve-imports ((# # license:) (#) (#) (#) ...)]
> 2867: 1 [resolve-interface (gnu packages javascript) #:select ...]
> In unknown file:
>    ?: 0 [scm-error misc-error #f ...]
> 
> ERROR: In procedure scm-error:
> ERROR: no code for module (gnu packages javascript)
> builder for `/gnu/store/58ji3i7a0dl2sgf39m16l90alnyyvc9j-guix-latest.drv' failed with exit code 1
> guix pull: error: build failed: build of `/gnu/store/58ji3i7a0dl2sgf39m16l90alnyyvc9j-guix-latest.drv' failed
> ===========================================================================
> 
> Looking at the patch, I think the problem is in
> `#:use-module (gnu packages javascript)'.

An embarrassing mistake :/

I still had gnu/packages/javascript.go in my source tree, so I didn't
notice the problem until I tried building Guix on another machine.

I ended up doing `make clean-go && rm gnu/packages/*.go`, and I was
surprised by how many old modules I had lying around. I will get into
the habit of doing this more often.