Clojure/SLIME/Emacs questions

Stefan Tilkov

unread,

Dec 31, 2009, 6:15:58 AM12/31/09

to clo...@googlegroups.com

Two quick Emacs/Clojure questions I can't seem to find the answer to:

- In his screencasts, Sean Devlin moves the mouse over an item in his REPL history and it becomes highlighted (and he can paste it to the current prompt with one click)
- Also in his screencasts, Lau Jensen types something like g-i-v, then hits some key combination and gets an expansion like get-integer-value

I can't seem to find out how to get these to work, any hint would be appreciated.

Thanks,
Stefan

--
Stefan Tilkov, http://www.innoq.com/blog/st/

Stefan Tilkov

unread,

Dec 31, 2009, 6:41:44 AM12/31/09

to clo...@googlegroups.com

I probably should have mentioned that I have Emacs/SLIME/Clojure working together nicely already, it's particularly these features that are missing.

Stefan

> --
> You received this message because you are subscribed to the Google
> Groups "Clojure" group.
> To post to this group, send email to clo...@googlegroups.com
> Note that posts from new members are moderated - please be patient with your first post.
> To unsubscribe from this group, send email to
> clojure+u...@googlegroups.com
> For more options, visit this group at
> http://groups.google.com/group/clojure?hl=en

Baishampayan Ghose

unread,

Dec 31, 2009, 7:07:41 AM12/31/09

to clo...@googlegroups.com

> - Also in his screencasts, Lau Jensen types something like g-i-v,
> then hits some key combination and gets an expansion like
> get-integer-value

This one is called fuzzy-completion, which for some unknown reasons
doesn't work with SLIME + Clojure.

It used to work perfectly with Common Lisp. I would appreciate a
solution to this too.

Regards,
BG

--
Baishampayan Ghose <b.g...@ocricket.com>
oCricket.com

Steve Purcell

unread,

Dec 31, 2009, 7:19:49 AM12/31/09

to clo...@googlegroups.com

Well, it clearly works for Lau, but then he says in one of the screencasts that he's using an old version of Slime; if you install technomancy's slime package from ELPA, I believe you don't get all the slime extensions, which would easily explain why the fuzzy completion doesn't work. I'm using that slime from ELPA, and fuzzy completion is also not working for me.

When the fuzzy completion extension *is* available, then the following should work to enable it:

(eval-after-load "slime"
'(progn
(require 'slime-fuzzy)
(setq slime-complete-symbol*-fancy t)
(setq slime-complete-symbol-function 'slime-fuzzy-complete-symbol)))

-Steve

Joost

unread,

Dec 31, 2009, 9:38:00 AM12/31/09

to Clojure

On 31 dec, 13:07, Baishampayan Ghose <b.gh...@ocricket.com> wrote:
>
> This one is called fuzzy-completion, which for some unknown reasons
> doesn't work with SLIME + Clojure.
>
> It used to work perfectly with Common Lisp. I would appreciate a
> solution to this too.

Works for me, but you must have slime-fuzzy or slime-fancy enabled. So
you need something like this in your emacs config:

(slime-setup
'(slime-fancy ;; turns on fancy inspector, autodoc and other useful
stuff
slime-highlight-edits))

This binds (slime-fuzzy-complete-symbol) to C-c M-i

You can also use hippie-expand, which does the same sort of
completion, but doesn't query the running lisp.

Joost.

Sean Devlin

unread,

Dec 31, 2009, 10:39:13 AM12/31/09

to Clojure

Stefan,
I run OSX, so I use command-click to bring the item to my REPL. I
imagine a right click would do the job in other platforms.

Sean

Stefan Tilkov

unread,

Dec 31, 2009, 11:07:13 AM12/31/09

to clo...@googlegroups.com

On Dec 31, 2009, at 4:39 PM, Sean Devlin wrote:

> Stefan,
> I run OSX, so I use command-click to bring the item to my REPL. I
> imagine a right click would do the job in other platforms.
>

Thanks Sean, but I am on OS X too – the line simply doesn't get highlighted, so I assumed there must be some difference in our setup.

Best,
Stefan

> Sean
>
> On Dec 31, 6:15 am, Stefan Tilkov <stefan.til...@innoq.com> wrote:
>> Two quick Emacs/Clojure questions I can't seem to find the answer to:
>>
>> - In his screencasts, Sean Devlin moves the mouse over an item in his REPL history and it becomes highlighted (and he can paste it to the current prompt with one click)
>> - Also in his screencasts, Lau Jensen types something like g-i-v, then hits some key combination and gets an expansion like get-integer-value
>>
>> I can't seem to find out how to get these to work, any hint would be appreciated.
>>
>> Thanks,
>> Stefan
>>
>> --
>> Stefan Tilkov,http://www.innoq.com/blog/st/
>

william douglas

unread,

Dec 31, 2009, 11:44:04 AM12/31/09

to clojure

> Well, it clearly works for Lau, but then he says in one of the screencasts that he's using an old version of Slime; if you install technomancy's slime package from > ELPA, I believe you don't get all the slime extensions, which would easily explain why the fuzzy completion doesn't work. I'm using that slime from ELPA, and > > fuzzy completion is also not working for me.

Right now the ELPA slime package does not include contrib (which
contains slime-fuzzy and a number of other things). I am attempting
to add just the required pieces to ELPA to get slime-fuzzy working.
If you would like fuzzy complete to work before this, you will want to
grab slime from technomancy's github repo and then add something like
the following to your .emacs:

(add-to-list 'load-path "~/slime/contrib/")

(eval-after-load "slime"
'(progn
(require 'slime-fuzzy)
(setq slime-complete-symbol*-fancy t)
(setq slime-complete-symbol-function 'slime-fuzzy-complete-symbol)))

Note that if you are using slime-connect the fuzzy complete might not
work though the following patch may fix that for you if you are using
lein swank to launch your swank session (patch from Takeshi Banse,
tak...@laafc.net):

diff --git a/src/leiningen/core.clj b/src/leiningen/core.clj
index a5261f4..685aab1 100644
--- a/src/leiningen/core.clj
+++ b/src/leiningen/core.clj
@@ -81,4 +81,5 @@ (defn -main [& [task & args]]
(abort (format "Wrong number of arguments to task %s."
task)))))
;; In case tests or some other task started any:
- (shutdown-agents)))
+ ;;(shutdown-agents)
+ ))

Rob Wolfe

unread,

Dec 31, 2009, 1:15:54 PM12/31/09

to clo...@googlegroups.com

Stefan Tilkov <stefan...@innoq.com> writes:

> Two quick Emacs/Clojure questions I can't seem to find the answer to:
>
> - In his screencasts, Sean Devlin moves the mouse over an item in his
> REPL history and it becomes highlighted (and he can paste it to the
> current prompt with one click)
> - Also in his screencasts, Lau Jensen types something like g-i-v, then hits some key combination and gets an expansion like get-integer-value
>
> I can't seem to find out how to get these to work, any hint would be appreciated.

I did it like this (I assume that clojure-mode.el has been installed):

1. downloaded slime and swank-clojure

$ pwd
/home/rw/opt/slime
$ head -n 1 .git/remotes/origin
URL: git://git.boinkor.net/slime.git
$ git pull
Already up-to-date.

$ pwd
/home/rw/opt/swank-clojure
$ head -n 1 .git/remotes/origin
URL: git://github.com/jochu/swank-clojure.git
$ git pull
Already up-to-date.

2. configured Emacs
(version: GNU Emacs 23.1.50.2 (i686-pc-linux-gnu, GTK+ Version 2.12.11))

<.emacs>

(add-to-list 'load-path "~/opt/slime/")
(add-to-list 'load-path "~/opt/slime/contrib/")
(add-to-list 'load-path "~/opt/swank-clojure/src/emacs")

(require 'clojure-mode)
(setq auto-mode-alist
(cons '("\\.clj$" . clojure-mode) auto-mode-alist))

(require 'swank-clojure-autoload)
(swank-clojure-config
(setq swank-clojure-jar-path "~/opt/clojure/clojure-git.jar")
(setq swank-clojure-extra-classpaths (list
"~/opt/clojure/jline.jar"
"~/src/clojure/gnu/clojure/clojure/src/jvm"))
(setq swank-clojure-extra-vm-args
'("-agentlib:jdwp=transport=dt_socket,address=8888,server=y,suspend=n")))

(require 'slime)
(slime-setup '(slime-repl))

(add-hook 'clojure-mode-hook
'(lambda ()
(define-key clojure-mode-map "\C-cc" 'comment-region)
(define-key clojure-mode-map "\C-cu" 'uncomment-region)))

(eval-after-load "slime"
'(progn
(require 'slime-fuzzy)
(setq slime-complete-symbol*-fancy t)
(setq slime-complete-symbol-function 'slime-fuzzy-complete-symbol)))

(defun run-clojure ()
"Starts clojure in Slime"
(interactive)
(slime 'clojure))

</.emacs>

And now "M-x run-clojure" starts slime with working completion.
For example after writing "w-o" and hitting TAB key I can see this:

user> (w-o
Completion: Flags: Score:
------------------------------ -------- --------
with-open -f---m-- 20,43
with-out-str -f---m-- 20,00
[...]

HTH,
Rob

Stefan Tilkov

unread,

Dec 31, 2009, 1:17:36 PM12/31/09

to clo...@googlegroups.com

Thanks for the quick response to everyone.

I downloaded slime-fuzzy.el from here:

http://elder-gods.org/~larry/repos/slime-tracker/contrib/slime-fuzzy.el

, put it into ~/.emacs.d/misc and then added this to my ~/.emacs:

(setq load-path (cons "~/.emacs.d/misc" load-path))

(eval-after-load "slime"
'(progn
(require 'slime-fuzzy)
(setq slime-complete-symbol*-fancy t)
(setq slime-complete-symbol-function 'slime-fuzzy-complete-symbol)))

… and everything works perfectly.

Stefan

Steven E. Harris

unread,

Dec 31, 2009, 4:19:07 PM12/31/09

to clo...@googlegroups.com

Rob Wolfe <r...@smsnet.pl> writes:

> I did it like this (I assume that clojure-mode.el has been installed):

Thanks, Rob. I followed your instructions after not being able to get
Clojure to cooperate with my normal Swank/SLIME installation, and it
works, but only to a point: The SLIME REPL buffer (invoked via the
function `slime-repl') falls out of step with the *inferior-lisp*
buffer.

This never happens to me when using SLIME with a Common Lisp, such as
CLISP. Here, Clojure starts up just fine in the *inferior-lisp* buffer,
but the SLIME REPL doesn't seem to connect properly. Maybe it connects
at the network level, but each form entered in the SLIME REPL yields
messages like this in the minibuffer:

,----
| ; pipelined request... (swank:listener-eval "
`----

That's a sign that the REPL is sending commands to the SWANK back-end,
but it's not receiving a timely response. Meanwhile, the Clojure REPL in
the *inferior-lisp* buffer remains responsive.

Trying to evaluate Clojure forms from a buffer other than the REPL (via,
say, `C-x e', usually doesn't work. There's no error reported, but none
of the resulting values display in the minibuffer, and any definitions
evaluated there have no apparent side effect on the Clojure REPL
available in the *inferior-lisp* buffer.

I've also noticed that the SLIME "presentations" feature doesn't work
with Clojure. That may be a symptom of the SLIME REPL not meeting up
properly with the Clojure REPL.

--
Steven E. Harris

Stefan Kamphausen

unread,

Dec 31, 2009, 8:17:38 PM12/31/09

to Clojure

Hi,

do you use a rather recent checkout of SLIME? If so, you may want to
read the thread http://groups.google.com/group/clojure/browse_frm/thread/3e5f416e3f2a1884/337057edae5dcdc3

Can't say whether that's related to your problem, though.

Cheers,
Stefan

Steve Purcell

unread,

Dec 31, 2009, 12:25:43 PM12/31/09

to clo...@googlegroups.com

Indeed -- that works nicely. I tried going back to a completely non-ELPA-ized setup, but it was too painful; the trick was installing technomancy's github repo of slime *in addition* to the ELPA packages, which all depend on each other.

-Steve

Stefan Kamphausen

unread,

Jan 1, 2010, 6:30:21 AM1/1/10

to Clojure

FWIW I have never touched ELPA but got a setup with SLIME from CVS
plus SBCL and Clojure. A rudimentary description can be found at
http://www.skamphausen.de/cgi-bin/ska/My_Clojure_Setup

The only part that feels tricky is compiling swank for which I needed
a Maven setup, IIRC.

Regards,
Stefan

Edmund Jackson

unread,

Jan 1, 2010, 8:16:15 AM1/1/10

to clo...@googlegroups.com

and FWIW I got the whole thing going without ELPA by using the
instructions here:

http://learnclojure.blogspot.com/2009/11/installing-clojure-on-ubuntu-910-karmic.html

but on OSX.

> --
> You received this message because you are subscribed to the Google
> Groups "Clojure" group.
> To post to this group, send email to clo...@googlegroups.com
> Note that posts from new members are moderated - please be patient
> with your first post.
> To unsubscribe from this group, send email to
> clojure+u...@googlegroups.com
> For more options, visit this group at
> http://groups.google.com/group/clojure?hl=en

Edmund

"The future is here. It's just not widely distributed yet"
-- Gibson

Konrad Hinsen

unread,

Jan 1, 2010, 10:35:59 AM1/1/10

to clo...@googlegroups.com

I just run swank from source code, uncompiled. No Maven, no worry :-)

Konrad.

Stefan Kamphausen

unread,

Jan 1, 2010, 11:29:33 AM1/1/10

to Clojure

Hi,

I had some errors before compiling and putting the resulting JAR on
the CP. But I'm not much of a Java-hero. What is your classpath? I
have swank-clojure/target/swank-clojure-1.0-SNAPSHOT.jar on it.

Cheers,
Stefan

Konrad Hinsen

unread,

Jan 1, 2010, 11:46:39 AM1/1/10

to clo...@googlegroups.com

On 01.01.2010, at 17:29, Stefan Kamphausen wrote:

> I had some errors before compiling and putting the resulting JAR on
> the CP. But I'm not much of a Java-hero. What is your classpath? I
> have swank-clojure/target/swank-clojure-1.0-SNAPSHOT.jar on it.

I always work with the latest Clojure ("new" branch from github).
Here's my swank setup in .emacs:

(setq swank-clojure-classpath
(list "~/Development/clojure/clojure.jar"
"~/Development/clojure-contrib/clojure-contrib.jar"
"~Development/clojure-dev-utilities"))

The last entry is a directory which contains among other stuff the
directory "swank" that contains swank.clj and the rest of the swank
source code.

Konrad.

Steven E. Harris

unread,

Jan 1, 2010, 12:19:11 PM1/1/10

to clo...@googlegroups.com

Stefan Kamphausen <ska...@googlemail.com> writes:

> you may want to read the thread
> http://groups.google.com/group/clojure/browse_frm/thread/3e5f416e3f2a1884/337057edae5dcdc3

I had read that thread /twice/ before, as well as the thread on the
SLIME mailing list, but I had neglected to try disabling
autodoc-mode. This morning, doing so makes things work much better.

However, I still notice that when I first run SLIME, the *inferior-lisp*
buffer comes up with a responsive REPL, but the SLIME REPL connection is
flaky. It takes a long time to finally come through, if it ever
does. Perhaps there's some race condition there.

Sometimes taking focus away from the Emacs window, doing something,
else, and returning to Emacs will cause the SLIME REPL to finally wake
up and finish its connection sequence. At that point the animated REPL
banner flies out and I have a working REPL.

But presentations still don't work.

--
Steven E. Harris

Steven E. Harris

unread,

Jan 1, 2010, 12:36:18 PM1/1/10

to clo...@googlegroups.com

"Steven E. Harris" <s...@panix.com> writes:

> This morning, doing so makes things work much better.

Still I see this in the *inferior-lisp* buffer when starting slime:

,----
| (require 'swank.swank)
|
| (swank.swank/ignore-protocol-version "2009-12-23")
|
| (swank.swank/start-server "c:/DOCUME~1/seh/LOCALS~1/Temp/slime.3420" :encoding "iso-latin-1-unix")
|
| Listening for transport dt_socket at address: 8888
| Clojure 1.1.0
| user=>
| WARNING: reader macro ^ is deprecated; use meta instead
| WARNING: reader macro ^ is deprecated; use meta instead
| WARNING: reader macro ^ is deprecated; use meta instead
| WARNING: reader macro ^ is deprecated; use meta instead
| WARNING: reader macro ^ is deprecated; use meta instead
| WARNING: reader macro ^ is deprecated; use meta instead
| WARNING: reader macro ^ is deprecated; use meta instead
| WARNING: reader macro ^ is deprecated; use meta instead
| WARNING: reader macro ^ is deprecated; use meta instead
| nil
| user=> user=> "2009-12-23"
| user=> user=> Connection opened on local port 3464
| #<ServerSocket ServerSocket[addr=0.0.0.0/0.0.0.0,port=0,localport=3464]>
| user=> user=> user=>
`----

Note that I'm using Clojure 1.1.0. Could that be the problem?
This is within GNU Emacs 23.1.1 on Windows XP.

Once that *inferior-lisp* buffer is up, if I run `slime-repl', the REPL
may take an arbitrarily long time to finally connect to Swank. And if it
does ever connect, and I shut it down via the "sayoonara" command, I
can't get it to start again successfully.

--
Steven E. Harris

Rob Wolfe

unread,

Jan 1, 2010, 1:18:13 PM1/1/10

to clo...@googlegroups.com

"Steven E. Harris" <s...@panix.com> writes:

> Rob Wolfe <r...@smsnet.pl> writes:
>
>> I did it like this (I assume that clojure-mode.el has been installed):
>
> Thanks, Rob. I followed your instructions after not being able to get
> Clojure to cooperate with my normal Swank/SLIME installation, and it
> works, but only to a point: The SLIME REPL buffer (invoked via the
> function `slime-repl') falls out of step with the *inferior-lisp*
> buffer.

So how do you exactly start SLIME for Clojure?
These settings:

(require 'slime)
(slime-setup '(slime-repl))

(defun run-clojure ()
"Starts clojure in Slime"
(interactive)
(slime 'clojure))

should cause automatic start of *inferior-lisp* and *slime-repl clojure*
after typing "M-x run-clojure"

>
> This never happens to me when using SLIME with a Common Lisp, such as
> CLISP. Here, Clojure starts up just fine in the *inferior-lisp* buffer,
> but the SLIME REPL doesn't seem to connect properly. Maybe it connects
> at the network level, but each form entered in the SLIME REPL yields
> messages like this in the minibuffer:
>
> ,----
> | ; pipelined request... (swank:listener-eval "
> `----

What do you see in *messages* buffer? Maybe there are some helpful
messages. I can see this:
"""
Polling "/tmp/slime.6251".. (Abort with `M-x slime-abort-connection'.) [31 times]
Connecting to Swank on port 40585.. [2 times]
Connected. Hacks and glory await!
"""

What version of Clojure do you use? I use clojure.jar built from sources
taken from GIT repository (I tried Clojure 1.0 and it worked too).

I agree with Stefan Kamphausen that "connection-serve" in swank.clj
should not fail silently. Now we don't know what is the problem.
If I were you I would change this function as Stefan suggested
and see what is going on:

<code>
(defn- connection-serve [conn]
(let [control
(dothread-swank
(thread-set-name "Swank Control Thread")
(try
(control-loop conn)
(catch Exception e
;; fail silently
(.printStackTrace e)
(.printStackTrace (.getCause e))
nil)))
read
(dothread-swank
(thread-set-name "Read Loop Thread")
(try
(read-loop conn control)
(catch Exception e
;; This could be put somewhere better
(.interrupt control)
(dosync (alter *connections* (partial remove #{conn}))))))]
(dosync
(ref-set (conn :control-thread) control)
(ref-set (conn :read-thread) read))))
</code>

>
> That's a sign that the REPL is sending commands to the SWANK back-end,
> but it's not receiving a timely response. Meanwhile, the Clojure REPL in
> the *inferior-lisp* buffer remains responsive.
>
> Trying to evaluate Clojure forms from a buffer other than the REPL (via,
> say, `C-x e', usually doesn't work. There's no error reported, but none
> of the resulting values display in the minibuffer, and any definitions
> evaluated there have no apparent side effect on the Clojure REPL
> available in the *inferior-lisp* buffer.
>
> I've also noticed that the SLIME "presentations" feature doesn't work
> with Clojure. That may be a symptom of the SLIME REPL not meeting up
> properly with the Clojure REPL.

Yes, apparently connection with *slime-repl clojure* failed
and it does not work properly now.

HTH,
Rob

Steven E. Harris

unread,

Jan 1, 2010, 2:06:08 PM1/1/10

to clo...@googlegroups.com

Rob Wolfe <r...@smsnet.pl> writes:

> So how do you exactly start SLIME for Clojure?

I've tried both `C-u - M-x slime RET clojure' and `run-clojure', which
both amount to the same thing eventually

> These settings:

[...]

> should cause automatic start of *inferior-lisp* and *slime-repl clojure*
> after typing "M-x run-clojure"

They do -- sometimes -- /eventually/. There is a very long gap between
the *inferior-lisp* process starting and the SLIME-Swank connection
completing, and the REPL appears only after that connection succeeds.

> What do you see in *messages* buffer?

The same as yours:

,----[ *Messages* buffer ]
| Polling "c:/DOCUME~1/seh/LOCALS~1/Temp/slime.5844"..
| (Abort with `M-x slime-abort-connection'.) [22 times]
| Connecting to Swank on port 3681.. [2 times]
| Connected. Let the hacking commence!
`----

But note that the connection attempt message repeated 22 times. And in
between the "connecting" and "connected" messages, there was a lag of
maybe a minute, during which I went in and interacted with the Clojure
REPL in the *inferior-lisp* buffer, trying to kick the connection to
completion.

If I just start SLIME and /don't/ interact with the *inferior-lisp*
buffer, the connection "never" completes, or at least not within the
five minutes I let it sit undisturbed.

This is the only content in the *slime-events* buffer while I'm waiting
for the connection to complete:

,----[ *slime-events* buffer ]
| (:emacs-rex
| (swank:connection-info)
| "COMMON-LISP-USER" t 1)
`----

> What version of Clojure do you use?

The Clojure 1.1.0 Jar that I downloaded from code.google.com.

> I use clojure.jar built from sources taken from GIT repository (I
> tried Clojure 1.0 and it worked too).

That's a good sign that something more recent than Clojure 1.0 should
work.

> I agree with Stefan Kamphausen that "connection-serve" in swank.clj
> should not fail silently. Now we don't know what is the problem.

I changed `connection-serve' per your suggestion.
The Swank back-end didn't dump any exception traces to the
*inferior-lisp* buffer.

The behavior looks like it suffers a race, or maybe the socket
connection request isn't noticed as intended.

--
Steven E. Harris

Steven E. Harris

unread,

Jan 1, 2010, 2:55:51 PM1/1/10

to clo...@googlegroups.com

"Steven E. Harris" <s...@panix.com> writes:

> This is the only content in the *slime-events* buffer while I'm waiting
> for the connection to complete:
>
> ,----[ *slime-events* buffer ]
> | (:emacs-rex
> | (swank:connection-info)
> | "COMMON-LISP-USER" t 1)
> `----

I also confirmed that `slime-set-connection-info' isn't being called
until after the long lag, which means that the REX above is part of
what's hanging. Either the Swank side isn't receiving the request or
it's not responding from the SLIME function `connection-info' in
file basic.clj.

The hang can be well /before/ this point too, but I've at least
established that it's not after `slime-set-connection-info'.

My next step will be to instrument the Swank side to see when
`connection-info' is being called.

--
Steven E. Harris

Phil Hagelberg

unread,

Jan 1, 2010, 4:45:13 PM1/1/10

to clo...@googlegroups.com

Rob Wolfe <r...@smsnet.pl> writes:

> Stefan Tilkov <stefan...@innoq.com> writes:
>
>> Two quick Emacs/Clojure questions I can't seem to find the answer to:
>>
>> - In his screencasts, Sean Devlin moves the mouse over an item in his
>> REPL history and it becomes highlighted (and he can paste it to the
>> current prompt with one click)
>> - Also in his screencasts, Lau Jensen types something like g-i-v, then hits some key combination and gets an expansion like get-integer-value
>>
>> I can't seem to find out how to get these to work, any hint would be appreciated.
>
> I did it like this (I assume that clojure-mode.el has been installed):
>
> 1. downloaded slime and swank-clojure

You can get this to work, but as you can see it's very manual and
error-prone. The basic slime REPL is working using automated
installation from ELPA, and William Douglas is working on adding fuzzy
completion to that. It's highly recommended that you follow this
approach rather than copying and pasting several pages of one-off
configuration. For instance, adding jline to your classpath is redundant
if you're using slime.

-Phil

Steven E. Harris

unread,

Jan 1, 2010, 8:15:27 PM1/1/10

to clo...@googlegroups.com

"Steven E. Harris" <s...@panix.com> writes:

> My next step will be to instrument the Swank side to see when
> `connection-info' is being called.

I found the problem: swank.util.sys/get-pid.

It looks like the JMX call to get the PID hangs. If I visit the
*inferior-lisp* buffer's REPL and evaluate the following form, the REPL
finishes connecting immediately:

,----[ Fragment of `get-pid' function ]
| (.. java.lang.management.ManagementFactory
| (getRuntimeMXBean) (getName) (split "@"))
`----

Apparently forcing more than one call through that path kicks it out of
some wait, and the thread blocked on the Swank-side function
`swank.commands.basic/connection-info' finally finishes its call to
get-pid.

Note that calling

java.lang.management.ManagementFactory#getRuntimeMXBean()

is not sufficient to unblock the thread in `connection-info'. It takes
the following call to

RuntimeMXBean#getName()

to unblock the wait.

Is there a separate mailing list to which I should report this problem?

--
Steven E. Harris

Phil Hagelberg

unread,

Jan 2, 2010, 12:20:50 AM1/2/10

to clo...@googlegroups.com

"Steven E. Harris" <s...@panix.com> writes:

> Is there a separate mailing list to which I should report this problem?

There's http://groups.google.com/group/swank-clojure, though for this
particular issue you can just continue in this thread.

But I'm not quite following on the exact problem and specific repro
case; could you clarify? Does it occur with the latest version of
swank-clojure?

-Phil

Rob Wolfe

unread,

Jan 2, 2010, 5:11:54 AM1/2/10

to clo...@googlegroups.com

Phil Hagelberg <ph...@hagelb.org> writes:

> Rob Wolfe <r...@smsnet.pl> writes:
>
>> Stefan Tilkov <stefan...@innoq.com> writes:
>>
>>> Two quick Emacs/Clojure questions I can't seem to find the answer to:
>>>
>>> - In his screencasts, Sean Devlin moves the mouse over an item in his
>>> REPL history and it becomes highlighted (and he can paste it to the
>>> current prompt with one click)
>>> - Also in his screencasts, Lau Jensen types something like g-i-v, then hits some key combination and gets an expansion like get-integer-value
>>>
>>> I can't seem to find out how to get these to work, any hint would be appreciated.
>>
>> I did it like this (I assume that clojure-mode.el has been installed):
>>
>> 1. downloaded slime and swank-clojure
>
> You can get this to work, but as you can see it's very manual and
> error-prone. The basic slime REPL is working using automated

The question was not about "basic" slime REPL.

> installation from ELPA, and William Douglas is working on adding fuzzy
> completion to that. It's highly recommended that you follow this

I will, if it will work.

> approach rather than copying and pasting several pages of one-off
> configuration. For instance, adding jline to your classpath is redundant
> if you're using slime.

Where do you see "several pages"? I don't like this error-prone manual
work, but so far only this work for me, that's why I showed this.
Sorry if it is a pain for you to read.

Rob

Steven E. Harris

unread,

Jan 2, 2010, 9:45:11 AM1/2/10

to clo...@googlegroups.com

Phil Hagelberg <ph...@hagelb.org> writes:

> But I'm not quite following on the exact problem and specific repro
> case; could you clarify?

Yes. When I start SLIME, the Swank back-end starts and presents the
Clojure REPL in the *inferior-lisp* buffer. Right after that, the SLIME
front-end sends an "Emacs REX" command to the Swank back-end, asking it
to invoke the `connection-info' function.

At this point, the SLIME front-end has reported that it's trying to
connect. The Swank back-end receives the REX request, invokes the
`connection-info' method on some thread separate from the thread
servicing the Clojure REPL, and hangs in its call to `get-pid'.

Because the Swank back-end thread servicing the call to
`connection-info' hasn't responded, the SLIME front-end is hung waiting
for its REX to complete to gather the `connection-info' result. Not
until the SLIME front-end has received that result will it finally
announce that it succeeded connecting and present the SLIME REPL.

It's not just the SLIME REPL that won't work until that
`connection-info' REX completes. Any buffers that attempt to submit
forms to Clojure for evaluation will not receive a result; the requests
are merely queued because the SLIME front-end has not completed its
connection and initial handshake.

The call to `get-pid' hangs on its call to

RuntimeMXBean#getName()

which is part of the first path taken by `get-pid':

,----

| (.. java.lang.management.ManagementFactory
| (getRuntimeMXBean)
| (getName)
| (split "@"))
`----

Now, I found a way to unblock that hung call to RuntimeMXBean#getName():
In the Clojure REPL available in the *inferior-lisp* buffer, I can
evaluate the following form:

,----
| (.. java.lang.management.ManagementFactory (getRuntimeMXBean) (getName)
`----

That evaluation takes place in the thread servicing the Clojure REPL,
which is a different thread from the one blocked in the REX call to
`connection-info' and, transitively, `get-pid'.

The above form evaluates immediately in he Clojure REPL and, as a
side-effect I don't yet understand, /also/ causes the call blocked in
the Swank REX-servicing thread to complete as well. Hence, `get-pid'
completes, the `connection-info' completes, then the SLIME front-end
receives its reply and finally sets up the SLIME REPL. From then on, all
is well.

I've been searching the Web looking for similar complaints about such a
call hanging, and so far I've come up blank. Again, note that calling
ManagementFactory#getRuntimeMXBean() from the Clojure REPL is not
sufficient to unblock the other thread hung in `get-pid'; it takes
calling RuntimeMXBean#get() -- actually using the RuntimeMXBean
instance -- to unblock the other thread. Perhaps there's some lazy
initialization going on in Sun's RuntimeMXBean implementation.

> Does it occur with the latest version of swank-clojure?

Yes. I'm tracking the tip of the "master" branch via Git.

Here's the stack:

o swank-clojure
Git head.

o SLIME
Both CVS head and git://git.boinkor.net/slime.git head behave the same
way, as the offending calls are in swank-clojure.

o Clojure 1.1.0
Downloaded Jar from code.google.com.

o GNU Emacs 23.1.1

o Windows XP 32-bit

--
Steven E. Harris

Steven E. Harris

unread,

Jan 2, 2010, 9:49:56 AM1/2/10

to clo...@googlegroups.com

"Steven E. Harris" <s...@panix.com> writes:

> it takes calling RuntimeMXBean#get() -- actually using the
> RuntimeMXBean instance -- to unblock the other thread.

I meant RuntimeMXBean#getName() there.
My typing is poor this morning.

--
Steven E. Harris

Rob Wolfe

unread,

Jan 2, 2010, 10:19:40 AM1/2/10

to clo...@googlegroups.com

"Steven E. Harris" <s...@panix.com> writes:

[...]

> Here's the stack:
>
> o swank-clojure
> Git head.
>
> o SLIME
> Both CVS head and git://git.boinkor.net/slime.git head behave the same
> way, as the offending calls are in swank-clojure.
>
> o Clojure 1.1.0
> Downloaded Jar from code.google.com.
>
> o GNU Emacs 23.1.1
>
> o Windows XP 32-bit

Unfortunately I have no idea what is the reason of this problem,
but I'm curious what Java version are you using?
Have you tried by any chance two different Java versions with the same
result?

user> (map #(System/getProperty %) ["java.version" "java.vendor" "java.vm.version"])
("1.5.0_17" "Sun Microsystems Inc." "1.5.0_17-b04")
user> (map #(System/getProperty %) ["os.name" "os.version" "os.arch"])
("Linux" "2.6.26-1-486" "i386")

Br,
Rob

Steven E. Harris

unread,

Jan 2, 2010, 10:49:16 AM1/2/10

to clo...@googlegroups.com

Rob Wolfe <r...@smsnet.pl> writes:

> Unfortunately I have no idea what is the reason of this problem,

That makes two of us.

> but I'm curious what Java version are you using?

user=> (map #(System/getProperty %)
["java.version" "java.vendor" "java.vm.version"])
("1.6.0_17" "Sun Microsystems Inc." "14.3-b01")
user=> (map #(System/getProperty %)
["os.name" "os.version" "os.arch"])
("Windows XP" "5.1" "x86")

> Have you tried by any chance two different Java versions with the same
> result?

No, that's the only one I have installed. In a few days when I return to
the office, I can try it against version 1.6.0_14 as well.

In the meantime, I'll continue the investigation.

--
Steven E. Harris

Steven E. Harris

unread,

Jan 2, 2010, 11:03:40 AM1/2/10

to clo...@googlegroups.com

"Steven E. Harris" <s...@panix.com> writes:

> In the meantime, I'll continue the investigation.

Here are two thread dumps of the Java process running Clojure and Swank,
which, at this time, has a live Clojure REPL, but the Swank call to
`connection-info' hasn't completed yet.

This first one comes via jvisualvm:

==========
2010-01-02 10:55:39
Full thread dump Java HotSpot(TM) Client VM (14.3-b01 mixed mode, sharing):

"RMI TCP Connection(4)-192.168.1.35" daemon prio=6 tid=0x02e43000 nid=0x15f8 runnable [0x0377f000]
java.lang.Thread.State: RUNNABLE
at java.net.SocketInputStream.socketRead0(Native Method)
at java.net.SocketInputStream.read(Unknown Source)
at java.io.BufferedInputStream.fill(Unknown Source)
at java.io.BufferedInputStream.read(Unknown Source)
- locked <0x23215700> (a java.io.BufferedInputStream)
at java.io.FilterInputStream.read(Unknown Source)
at sun.rmi.transport.tcp.TCPTransport.handleMessages(Unknown Source)
at sun.rmi.transport.tcp.TCPTransport$ConnectionHandler.run0(Unknown Source)
at sun.rmi.transport.tcp.TCPTransport$ConnectionHandler.run(Unknown Source)
at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(Unknown Source)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source)
at java.lang.Thread.run(Unknown Source)

Locked ownable synchronizers:
- <0x2321c928> (a java.util.concurrent.locks.ReentrantLock$NonfairSync)

"RMI TCP Connection(idle)" daemon prio=6 tid=0x02f74c00 nid=0x121c waiting on condition [0x0372f000]
java.lang.Thread.State: TIMED_WAITING (parking)
at sun.misc.Unsafe.park(Native Method)
- parking to wait for <0x231df420> (a java.util.concurrent.SynchronousQueue$TransferStack)
at java.util.concurrent.locks.LockSupport.parkNanos(Unknown Source)
at java.util.concurrent.SynchronousQueue$TransferStack.awaitFulfill(Unknown Source)
at java.util.concurrent.SynchronousQueue$TransferStack.transfer(Unknown Source)
at java.util.concurrent.SynchronousQueue.poll(Unknown Source)
at java.util.concurrent.ThreadPoolExecutor.getTask(Unknown Source)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source)
at java.lang.Thread.run(Unknown Source)