Freeze in term with latest emacs sources

[antoine-levitt]

Hi,

With emacs from latest git, running sml/setup and then M-x term freezes emacs. Might be related to https://lists.gnu.org/archive/html/bug-gnu-emacs/2016-10/msg00703.html.

[peterwvj]

I'm experiencing the same if I run M-x ansi-term.

Info:
smart-mode-line 20160818.508
GNU Emacs 26.0.50 (build 2, x86_64-pc-linux-gnu, GTK+ Version 3.18.9) of 2017-02-18

[peterwvj]

I'm experiencing the same if I run M-x ansi-term.

Info:
smart-mode-line 20160818.508
GNU Emacs 26.0.50 (build 2, x86_64-pc-linux-gnu, GTK+ Version 3.18.9) of 2017-02-18

[Mihara]

Same here.

For what it's worth, interrupting the frozen emacs with SIGUSR2 produces this debug output:

Debugger entered--entering a function:
* sml/-this-buffer-changed(1 5 0)
  file-accessible-directory-p("/home/mihara/.emacs.d/lisp/")
  cd-absolute("/home/mihara/.emacs.d/lisp")
  cd("/home/mihara/.emacs.d/lisp")
  #[(string) "�\306\232\203��\307\207�\310H\311U\2031�\312\313�\314#�\312\313�	T#��\314	O�\315�	T\nO!��\fB\211�,\207�\310H\316U\203@�\317�\314\320O!\207\307\207" [string first-colon second-colon filename fileline term-pending-frame "" t 0 26 string-match ":" 1 string-to-number 47 cd nil] 4]("//home/mihara/.emacs.d/lisp")
  ad-Advice-term-command-hook(#[(string) "�\306\232\203��\307\207�\310H\311U\2031�\312\313�\314#�\312\313�	T#��\314	O�\315�	T\nO!��\fB\211�,\207�\310H\316U\203@�\317�\314\320O!\207\307\207" [string first-colon second-colon filename fileline term-pending-frame "" t 0 26 string-match ":" 1 string-to-number 47 cd nil] 4] "//home/mihara/.emacs.d/lisp")
  apply(ad-Advice-term-command-hook #[(string) "�\306\232\203��\307\207�\310H\311U\2031�\312\313�\314#�\312\313�	T#��\314	O�\315�	T\nO!��\fB\211�,\207�\310H\316U\203@�\317�\314\320O!\207\307\207" [string first-colon second-colon filename fileline term-pending-frame "" t 0 26 string-match ":" 1 string-to-number 47 cd nil] 4] "//home/mihara/.emacs.d/lisp")
  #[128 "\300\301\302�#\207" [apply ad-Advice-term-command-hook #[(string) "�\306\232\203��\307\207�\310H\311U\2031�\312\313�\314#�\312\313�	T#��\314	O�\315�	T\nO!��\fB\211�,\207�\310H\316U\203@�\317�\314\320O!\207\307\207" [string first-colon second-colon filename fileline term-pending-frame "" t 0 26 string-match ":" 1 string-to-number 47 cd nil] 4] nil] 5 nil]("//home/mihara/.emacs.d/lisp")
  term-emulate-terminal(#<process *ansi-term*> "�//home/mihara/.emacs.d/lisp
\n")

Is there any hope someone more knowledgeable can get this sorted out? This is a real pain.

[Mihara]

Same here.

For what it's worth, interrupting the frozen emacs with SIGUSR2 produces this debug output:

Debugger entered--entering a function:
* sml/-this-buffer-changed(1 5 0)
  file-accessible-directory-p("/home/mihara/.emacs.d/lisp/")
  cd-absolute("/home/mihara/.emacs.d/lisp")
  cd("/home/mihara/.emacs.d/lisp")
  #[(string) "�\306\232\203��\307\207�\310H\311U\2031�\312\313�\314#�\312\313�	T#��\314	O�\315�	T\nO!��\fB\211�,\207�\310H\316U\203@�\317�\314\320O!\207\307\207" [string first-colon second-colon filename fileline term-pending-frame "" t 0 26 string-match ":" 1 string-to-number 47 cd nil] 4]("//home/mihara/.emacs.d/lisp")
  ad-Advice-term-command-hook(#[(string) "�\306\232\203��\307\207�\310H\311U\2031�\312\313�\314#�\312\313�	T#��\314	O�\315�	T\nO!��\fB\211�,\207�\310H\316U\203@�\317�\314\320O!\207\307\207" [string first-colon second-colon filename fileline term-pending-frame "" t 0 26 string-match ":" 1 string-to-number 47 cd nil] 4] "//home/mihara/.emacs.d/lisp")
  apply(ad-Advice-term-command-hook #[(string) "�\306\232\203��\307\207�\310H\311U\2031�\312\313�\314#�\312\313�	T#��\314	O�\315�	T\nO!��\fB\211�,\207�\310H\316U\203@�\317�\314\320O!\207\307\207" [string first-colon second-colon filename fileline term-pending-frame "" t 0 26 string-match ":" 1 string-to-number 47 cd nil] 4] "//home/mihara/.emacs.d/lisp")
  #[128 "\300\301\302�#\207" [apply ad-Advice-term-command-hook #[(string) "�\306\232\203��\307\207�\310H\311U\2031�\312\313�\314#�\312\313�	T#��\314	O�\315�	T\nO!��\fB\211�,\207�\310H\316U\203@�\317�\314\320O!\207\307\207" [string first-colon second-colon filename fileline term-pending-frame "" t 0 26 string-match ":" 1 string-to-number 47 cd nil] 4] nil] 5 nil]("//home/mihara/.emacs.d/lisp")
  term-emulate-terminal(#<process *ansi-term*> "�//home/mihara/.emacs.d/lisp
\n")

Is there any hope someone more knowledgeable can get this sorted out? This is a real pain.

[Malabarba]

Sorry for the silence. I'll try to have a look at it this week. It's weird that file-accessible-directory-p is triggering sml hooks.

[Malabarba]

Sorry for the silence. I'll try to have a look at it this week. It's weird that file-accessible-directory-p is triggering sml hooks.

[antoine-levitt]

For what it's worth, temporary workaround:

(ad-deactivate 'term-command-hook)
(ad-deactivate 'term-handle-ansi-terminal-messages)

[antoine-levitt]

For what it's worth, temporary workaround:

(ad-deactivate 'term-command-hook)
(ad-deactivate 'term-handle-ansi-terminal-messages)

[erreina]

Any further progress on this?

[erreina]

Any further progress on this?

[Malabarba]

No progress. The term advices used by sml are just meant to keep track of the shell's CWD. If it's causing hangs then I'm not opposed to simply disabling this feature.
Is this still a problem with latest Emacs?

[Malabarba]

No progress. The term advices used by sml are just meant to keep track of the shell's CWD. If it's causing hangs then I'm not opposed to simply disabling this feature.
Is this still a problem with latest Emacs?

[apauzies]

Yes, I have the issue with:

GNU Emacs 26.0.90 (build 1, x86_64-apple-darwin13.4.0, NS appkit-1265.21 Version 10.9.5 (Build 13F1911)) of 2017-10-11

Debugger entered--returning value: nil
  sml/-this-buffer-changed(1 5 0)
  cd-absolute("/Users/alex")
  cd("/Users/alex")
  #f(compiled-function (string) #<bytecode 0x40551ba3>)("//Users/alex")
  ad-Advice-term-command-hook(#f(compiled-function (string) #<bytecode 0x40551ba3>) "//Users/alex")
  apply(ad-Advice-term-command-hook #f(compiled-function (string) #<bytecode 0x40551ba3>) "//Users/alex")
  #f(advice-wrapper :around #f(compiled-function (string) #<bytecode 0x40551ba3>) ad-Advice-term-command-hook)("//Users/alex")
  term-emulate-terminal(#<process *ansi-term*> "\032//Users/alex\015\n\033[?1034h\033[01;35malex@imac\033[01;36m:~ $\033[00m ")

[apauzies]

Yes, I have the issue with:

GNU Emacs 26.0.90 (build 1, x86_64-apple-darwin13.4.0, NS appkit-1265.21 Version 10.9.5 (Build 13F1911)) of 2017-10-11

Debugger entered--returning value: nil
  sml/-this-buffer-changed(1 5 0)
  cd-absolute("/Users/alex")
  cd("/Users/alex")
  #f(compiled-function (string) #<bytecode 0x40551ba3>)("//Users/alex")
  ad-Advice-term-command-hook(#f(compiled-function (string) #<bytecode 0x40551ba3>) "//Users/alex")
  apply(ad-Advice-term-command-hook #f(compiled-function (string) #<bytecode 0x40551ba3>) "//Users/alex")
  #f(advice-wrapper :around #f(compiled-function (string) #<bytecode 0x40551ba3>) ad-Advice-term-command-hook)("//Users/alex")
  term-emulate-terminal(#<process *ansi-term*> "\032//Users/alex\015\n\033[?1034h\033[01;35malex@imac\033[01;36m:~ $\033[00m ")

[Malabarba]

Thanks for all the info.
I haven't been able to spend as much time as I'd like on package maintenance, so I welcome anyone who'd like to take a crack at this.
Meanwhile, I'll keep trying to find some time for it, and apologising for the delays.

[Malabarba]

Thanks for all the info.
I haven't been able to spend as much time as I'd like on package maintenance, so I welcome anyone who'd like to take a crack at this.
Meanwhile, I'll keep trying to find some time for it, and apologising for the delays.

[a-schaefers]

The workaround causes an annoying init warning message to pop up, I hope we can resolve this better

[a-schaefers]

The workaround causes an annoying init warning message to pop up, I hope we can resolve this better

[phil-s]

Edit: I missed that the commit was to a completely different repository, so the following isn't entirely applicable.

---

You shouldn't be deactivating all advice for these functions if you know that there are specific pieces of advice which are responsible -- who knows what else you've disabled.

Use ad-remove-advice for the specific advice, and then ad-activate to make it take effect.

(Of course, if the advice in question is added by this same library -- which I presume it is, given the lack of eval-after-load -- then the more sensible workaround would be to not define/activate that advice in the first place.)

[phil-s]

Edit: I missed that the commit was to a completely different repository, so the following isn't entirely applicable.

---

You shouldn't be deactivating all advice for these functions if you know that there are specific pieces of advice which are responsible -- who knows what else you've disabled.

Use ad-remove-advice for the specific advice, and then ad-activate to make it take effect.

(Of course, if the advice in question is added by this same library -- which I presume it is, given the lack of eval-after-load -- then the more sensible workaround would be to not define/activate that advice in the first place.)

[Malabarba]

Hi everyone.
I've removed both advices now. Let me know if it solves the issue as soon as Melpa rebuilds (I tried reproducing the issue here but it didn't happen for me).

[Malabarba]

Hi everyone.
I've removed both advices now. Let me know if it solves the issue as soon as Melpa rebuilds (I tried reproducing the issue here but it didn't happen for me).

[daviderestivo]

Hi Artur,

it works for me now.

Thanks and regards,
Davide

[daviderestivo]

Hi Artur,

it works for me now.

Thanks and regards,
Davide