Re: master 69fb12a: (loadhist-unload-element): Move ERT and cl-generic methods

Re: master 69fb12a: (loadhist-unload-element): Move ERT and cl-generic methods

[Alex]

monnier@iro.umontreal.ca (Stefan Monnier) writes:

> branch: master
> commit 69fb12a66b3d6b9bfb55d8bcd58bec2a8e7ca55b
> Author: Stefan Monnier <monnier@iro.umontreal.ca>
> Commit: Stefan Monnier <monnier@iro.umontreal.ca>
>
>     (loadhist-unload-element): Move ERT and cl-generic methods
>     
>     * lisp/loadhist.el (loadhist-unload-element): Don't define cl-generic
>     and ert methods here.
>     (loadhist-unload-element) <(head define-type)>: Remove unused var `slots'.
>     
>     * lisp/emacs-lisp/cl-generic.el (loadhist-unload-element): Define
>     unload method for cl-defmethod.
>     (cl-generic-ensure-function): Remove redundant `defalias'.
>     
>     * lisp/emacs-lisp/ert.el (ert-set-test): Move the current-load-list
>     setting here...
>     (ert-deftest): ...from here.
>     (loadhist-unload-element): Define unload method for ert-deftest.

This commit breaks defgeneric for me. Specifically, this:

(eieio--defgeneric-init-form 'test "Hello")

returns nil after this commit, leading to an assertion error in
eieio--defalias.

I found this out because I couldn't load Helm (the problem file is
helm-source.el).

Re: master 69fb12a: (loadhist-unload-element): Move ERT and cl-generic methods

[Stefan Monnier]

> This commit breaks defgeneric for me. Specifically, this:
> (eieio--defgeneric-init-form 'test "Hello")
> returns nil after this commit, leading to an assertion error in
> eieio--defalias.

Duh!  Thanks for the heads up.  Should be fixed now,


        Stefan

Re: master 69fb12a: (loadhist-unload-element): Move ERT and cl-generic methods

[Glenn Morris]

> branch: master
> commit 69fb12a66b3d6b9bfb55d8bcd58bec2a8e7ca55b
>
>     (loadhist-unload-element): Move ERT and cl-generic methods

It seems this causes the test
xref-elisp-test-find-defs-defgeneric-implicit-generic to fail.

Ref eg http://hydra.nixos.org/build/56999656

(It's harder than normal to figure these things out owing to hanging
tramp tests.)

Hanging Tramp tests (was: master 69fb12a: (loadhist-unload-element): Move ERT and cl-generic methods)

[Michael Albinus]

Glenn Morris <rgm@gnu.org> writes:

> (It's harder than normal to figure these things out owing to hanging
> tramp tests.)

I know :-(

I've nailed it down to the exact position the test hangs. Or not. It's
the form starting at line 3809 in tramp-tests.el.

I have *no* idea yet why it hangs sometimes. And only on hydra. If it
damages other people's tests too much, I will skip this test on hydra,
'tho I really would like to know what's up.

Comments?

Best regards, Michael.

Re: Hanging Tramp tests (was: master 69fb12a: (loadhist-unload-element): Move ERT and cl-generic methods)

[Noam Postavsky]

On Wed, Jul 26, 2017 at 2:06 PM, Michael Albinus <michael.albinus@gmx.de> wrote:
> Glenn Morris <rgm@gnu.org> writes:
>
>> (It's harder than normal to figure these things out owing to hanging
>> tramp tests.)
>
> I know :-(
>
> I've nailed it down to the exact position the test hangs. Or not. It's
> the form starting at line 3809 in tramp-tests.el.
>
> I have *no* idea yet why it hangs sometimes. And only on hydra. If it
> damages other people's tests too much, I will skip this test on hydra,
> 'tho I really would like to know what's up.
>
> Comments?

Maybe using an external process would be able to end the hang more
reliably than `with-timeout'?

    (start-process "*watchdog*" nil shell-file-name shell-command-switch
                   (format "sleep 350 ; kill -SIGUSR2 %d" (emacs-pid)))

Re: Hanging Tramp tests

[Michael Albinus]

Noam Postavsky <npostavs@users.sourceforge.net> writes:

Hi Noam,

> Maybe using an external process would be able to end the hang more
> reliably than `with-timeout'?
>
>     (start-process "*watchdog*" nil shell-file-name shell-command-switch
>                    (format "sleep 350 ; kill -SIGUSR2 %d" (emacs-pid)))

Done.

Best regards, Michael.

Re: master 69fb12a: (loadhist-unload-element): Move ERT and cl-generic methods

[Dmitry Gutov]

On 7/26/17 7:43 PM, Glenn Morris wrote:

> It seems this causes the test
> xref-elisp-test-find-defs-defgeneric-implicit-generic to fail.

Seems to work now, thanks. (Probably after 86c862767?)

Re: master 69fb12a: (loadhist-unload-element): Move ERT and cl-generic methods

[Stefan Monnier]

>> It seems this causes the test
>> xref-elisp-test-find-defs-defgeneric-implicit-generic to fail.
> Seems to work now, thanks. (Probably after 86c862767?)

Indeed,


        Stefan

Re: master 69fb12a: (loadhist-unload-element): Move ERT and cl-generic methods

[Rob Browning]

Glenn Morris <rgm@gnu.org> writes:

> (It's harder than normal to figure these things out owing to hanging
> tramp tests.)

In case it's helpful evidence, it looks like (at least) the 25.2 tramp
tests also hang when running under sbuild
(https://wiki.debian.org/sbuild).

-- 
Rob Browning
rlb @defaultvalue.org and @debian.org
GPG as of 2011-07-10 E6A9 DA3C C9FD 1FF8 C676 D2C4 C0F0 39E9 ED1B 597A
GPG as of 2002-11-03 14DD 432F AE39 534D B592 F9A0 25C8 D377 8C7E 73A4

Hanging Tramp tests (was: master 69fb12a: (loadhist-unload-element): Move ERT and cl-generic methods)

[Michael Albinus]

Rob Browning <rlb@defaultvalue.org> writes:

Hi Rob,

> In case it's helpful evidence, it looks like (at least) the 25.2 tramp
> tests also hang when running under sbuild

I'm not familiar with sbuild. Do you have tramp-tests.log for analysis?

Best regards, Michael.

Re: Hanging Tramp tests (was: master 69fb12a: (loadhist-unload-element): Move ERT and cl-generic methods)

[Rob Browning]

Michael Albinus <michael.albinus@gmx.de> writes:

> I'm not familiar with sbuild. Do you have tramp-tests.log for
> analysis?

Actually, spwhitton reminded me that it might be a problem with chroots
created by older versions of schroot, and sure enough, rebuilding the
chroot fixed it.

...which revealed a secondary issue.  It looks like (in 25.2, and
perhaps master too) elisp-mode-tests.el downcases the xref path
(location) for comparison in xref-elisp-test-run, but that seems to also
affect the filesystem lookup lower down (in xref--goto-location) such
that a handful of tests fail if the build path contains uppercase
characters.

In any case, I've hacked up a temporary fix by keeping the original as
"real-xref" and cloning that (and copying the location) to produce the
xref used by the modification and comparison.

Thanks
-- 
Rob Browning
rlb @defaultvalue.org and @debian.org
GPG as of 2011-07-10 E6A9 DA3C C9FD 1FF8 C676 D2C4 C0F0 39E9 ED1B 597A
GPG as of 2002-11-03 14DD 432F AE39 534D B592 F9A0 25C8 D377 8C7E 73A4

Re: Hanging Tramp tests

[Michael Albinus]

Rob Browning <rlb@defaultvalue.org> writes:

Hi Rob,

> ...which revealed a secondary issue.  It looks like (in 25.2, and
> perhaps master too) elisp-mode-tests.el downcases the xref path
> (location) for comparison in xref-elisp-test-run, but that seems to also
> affect the filesystem lookup lower down (in xref--goto-location) such
> that a handful of tests fail if the build path contains uppercase
> characters.

I'm sorry but this is rather out of my scope.

> In any case, I've hacked up a temporary fix by keeping the original as
> "real-xref" and cloning that (and copying the location) to produce the
> xref used by the modification and comparison.

Nobody else has reacted. Maybe you write a bug report via `report-emacs-bug'?

> Thanks

Best regards, Michael.

Re: Hanging Tramp tests

[Rob Browning]

Michael Albinus <michael.albinus@gmx.de> writes:

> Nobody else has reacted. Maybe you write a bug report via `report-emacs-bug'?

Done.

Thanks
-- 
Rob Browning
rlb @defaultvalue.org and @debian.org
GPG as of 2011-07-10 E6A9 DA3C C9FD 1FF8 C676 D2C4 C0F0 39E9 ED1B 597A
GPG as of 2002-11-03 14DD 432F AE39 534D B592 F9A0 25C8 D377 8C7E 73A4