gmane.lisp.openmcl.devel

Re: Why canot CCL load libSDL* ?

Gary Byers — 2012-05-23T02:38:16

The function CCL:OPEN-SHARED-LIBRARY essentially just passes its
argument to a foreign function (#_dlopen on Unix-based systems) and
returns something based on what that function returns or signals an
error.  The lisp package involved in your game uses code in the CFFI
package to call CCL:OPEN-SHARED-LIBRARY for you; it apparently tries
to look for "libSDL.so" under that name and a few version-specific 
variants; it tells you that none of those attempts was succesful but
doesn't tell you what problems were encountered.

Calling

? (CCL:OPEN-SHARED-LIBRARY "/usr/local/lib/libSDL.so")

directly would likely tell you something about what the problem is.

If that fails (signals an error), the error message may explain why.  (The
error message is just what #_dlopen returns - or, more accurately, what #_dlopen
causes a function named #_dlerror to return; if the error message seems a bit
cryptic, there isn't much that CCL can do to make it more intelligible for you.)

Since the file exists, some likely reasons for

Why canot CCL load libSDL* ?

z_axis — 2012-05-22T03:14:45

When i quickload a game from https://github.com/sykopomp/common-worm, CCL  
reports:


But:
lrwxr-xr-x  1 root  wheel      16>ls -l /usr/local/lib/libSDL*
lrwxr-xr-x  1 root  wheel      16 12  3 09:23  
/usr/local/lib/libSDL-1.2.so< at > -> libSDL-1.2.so.11
-rwxr-xr-x  1 root  wheel  452662 12  3 09:23  
/usr/local/lib/libSDL-1.2.so.11*
-rw-r--r--  1 root  wheel  559164 12  3 09:23 /usr/local/lib/libSDL.a
-rwxr-xr-x  1 root  wheel    1170 12  3 09:23 /usr/local/lib/libSDL.la*
lrwxr-xr-x  1 root  wheel      16 12  3 09:23 /usr/local/lib/libSDL.so< at > ->  
libSDL-1.2.so.11
-rw-r--r--  1 root  wheel     872 12  3 09:23 /usr/local/lib/libSDLmain.a

FreeBSD mybsd.zsoft.com 9.0-RELEASE FreeBSD 9.0-RELEASE #0: Tue Jan  3  
07:15:25 UTC 2012      
root< at >obrian.cse.buffalo.edu:/usr/obj/usr/src/sys/GENERIC  i386



Sincerely!

Re: CXML-RNG, Plexippus XPath, and Xuriella XSLT

Steven Núñez — 2012-05-20T23:16:11

Interesting. Below is the output on my system (Darwin Kernel Version
11.3.0, MacOS C 10.7.3, Intel Core i7). Probably unrelated, but I notice
that QuickLisp isn't picking up the cxml-stp dependency, see below for
example of that.

I saw Dave pushed a change into GitHub, but following the thread, I
noticed it broke SBCL, so I'm not expecting it be fixed here, just
demonstrating the error I'm seeing.

Cheers,
- Steve


__________________________________________________________________
Welcome to Clozure Common Lisp Version 1.8-r15367M  (DarwinX8664)!
? (ql:quickload "xpath")
To load "xpath":
  Load 1 ASDF system:
    xpath
; Loading "xpath"
;;; Checking for wide character support... yes, using code points.
;;; Checking for wide character support... yes, using code points.
;;; Building Closure with CHARACTER RUNES
......
("xpath")
? (defparameter *document*
   (cxml:parse "<test a='1' b='2' xmlns:foo='http://foo'>
                          <child>hello world</child>
                          <foo:child>bar</f

Re: trunk may be unstable for a day or two

Gary Byers — 2012-05-21T10:53:29

The dust seems to have settled.

On Fri, 18 May 2012, Gary Byers wrote:

Re: CXML-RNG, Plexippus XPath, and Xuriella XSLT

Zach Beane — 2012-05-20T19:06:14


Git repo does not build in SBCL any more. I get this:

    There is no class named CXML-STP-IMPL::DOCUMENT.

Zach

Re: CXML-RNG, Plexippus XPath, and Xuriella XSLT

David Lichteblau — 2012-05-20T14:35:25

Hi,

Quoting Gary Byers (gb< at >clozure.com):
[...]

thanks, I've pushed a fix along these lines to cxml-stp git.  If you
(Stephen) are working with quicklisp, please keep in mind that, while
quicklisp provides a snapshot of cxml-stp from git, it won't update
immediately.  So please clone/pull from the git repo directly in this
case.

[...]

There is cl-libxml2, but (as its name suggests) it merely wraps a C
library.  (In return it's faster.)

If Plexippus doesn't work out of the box on CCL, it's because it doesn't
yet do non-trapping IEEE arithmetic on CCL, which means that the test
suite can't possibly pass, which in turn means that I consider CCL
unsupported.  I'll need it on CCL myself relatively soon, so you might
want to stay tuned -- but I'd also take patches :-).


d.

Re: CXML-RNG, Plexippus XPath, and Xuriella XSLT

2012-05-20T15:00:16

No problem here:

*-----------------------------------------------------------------------------
|  
| (ql:quickload "xpath")
| 
| (defvar *doc* 
|   (cxml:parse "<doc><a>foo</a><b>foo<a>Test</a></b></doc>"
| (cxml-dom:make-dom-builder))
|   "A simple xml document")
| 
| => *DOC*
| 
| (xpath:evaluate "/"  *doc*)
| 
| => #<XPATH:NODE-SET #<DOCUMENT {C8F9189}>, ... {B7DF609}>
| 
| (defvar all-a-nodes (xpath:evaluate "//a/text()"  *doc*))
| 
| => ALL-A-NODES
| 
| (xpath:do-node-set (a-node all-a-nodes)
|   (format t "Node ~A~%" (xpath:string-value a-node)))
| 
| Node foo
| Node Test
| => NIL
| 
*-------------------------------------------------------------------------------------


Maybe you can report to us your version/platform of CCL? And hoew did
you load cxml/plexippus etc.?

 Cheers, Ralf Mattes

Re: CXML-RNG, Plexippus XPath, and Xuriella XSLT

Gary Byers — 2012-05-20T14:09:05

This is just a guess - and  someone who has some idea of what the acronyms you
use mean may have a better one - but the code that you're trying to compile
may do something analogous to:

(defclass foo ()
  ())

(setf (find-class 'bar) (find-class 'foo))

... (typep x 'bar) ...

The spec says (in 4.3.7) that the proper name of a class is a valid
type specifier; in the code above, BAR isn't the proper name of any
class and likely isn't going to be recognized as a valid type
specifier in CCL; other implementations may effectively treat any use
of a symbol S (as a potential type-specifier) for which (FIND-CLASS S
NIL) returns a CLASS object C as being equivalent to C, even if
(CLASS-NAME C) returns something other than S.

You can persuade CCL to treat BAR as a valid type specifier in this
case by doing:

(deftype bar () 'foo)

before trying to use BAR as a type specifer.

[Of course, the problem you're having may be totally unrelated to the above.]



On Sun, 20 May 2012, Steven N?ez wrote:

CXML-RNG, Plexippus XPath, and Xuriella XSLT

Steven Núñez — 2012-05-20T10:54:58

Greetings,

Has anyone got any of the above working on CCL? Specifically I'm trying to use Plexippus, but get:

        Unknown type specifier: CXML-STP-IMPL::DOCUMENT-TYPE

When invoking any Plexippus functions. Not even the simple examples in the documentation run. This is on the latest 1.8 CCL.

Any other suggestions for XPath with CCL?

Regards,
- SteveN

--
Illation Pty Ltd
8/350 Collins Street
Melbourne 3000

T:   +61 3 8399 9442
M: +61 4 0096 4240

_______________________________________________
Openmcl-devel mailing list
Openmcl-devel< at >clozure.com
http://clozure.com/mailman/listinfo/openmcl-devel

trunk may be unstable for a day or two

Gary Byers — 2012-05-18T20:59:32

I want to check in a bunch of changes (around a dozen files, all in
the lisp-kernel directory) to the CCL trunk.  Most of the changes involve
low-level GC-related things; things seem to work (but could stand some
stress-testing) on x8664 Linux but may or may not even compile cleanly
on other platforms, and the easiest way for me to get things working
everywhere is to check the code in and start building and testing it.

People who use the trunk may want to refrain from doing an "svn update"
for the next day or two; I'll send an "all clear" message to this list
when the dust seems to have settled.

Re: Determining the Operating System on Windows

Michael Minerva — 2012-05-17T19:37:30

Hey Raymond,

Yes those do seem useful thank you!

Software type seems to always return: "Microsoft Windows" on Windows

and 

Software version will return: "6.1 Build 7600 (Workstation)" on Windows 7 and "5.1 Build 2600 (Workstation)" on Windows XP.  This corresponds with what I read about the win32 function "getVersionEx" which should return a value above 5.0 for XP and 6.1 for Windows 7.  

Thanks a lot,

--Mike
On May 17, 2012, at 12:38 PM, Raymond Wiker wrote:

Re: Determining the Operating System on Windows

Raymond Wiker — 2012-05-17T18:38:06


Do #'software-type and #'software-version return anything that is useful to you?

Alternatively, there may or may not be something useful on *features*...

Determining the Operating System on Windows

Michael Minerva — 2012-05-17T18:31:54

Hey Everyone,

I was wondering if there was an easy way in CCL to determine which operating system the application was running on.  More specifically I am trying to determine which version of Windows is running.  

Is there any easy way to do this with CCL or should I proceed into the realm of win32 hacking to get this done?

Thanks a bunch,

--Mike

DIRECTORY quibbles

Erik Pearson — 2012-05-16T17:26:58

I was tinkering with DIRECTORY, attempting to debug something with ASDF.

I discovered that (DIRECTORY "*") was returning directories in the current
working directory, but not directories with dots in their names.

Reading the CCL docs http://ccl.clozure.com/manual/chapter4.4.html (also
the docs included in trunk), the interface specifies a keyword option
:DIRECTORIES which should default to NIL. It is apparently defaulting to T,
and checking the source files it is coded this way.  I think the
documentation specifies the correct, or at least most conforming, behavior,
that is, to list files. Maybe it is less surprising to get directories by
default, by analogy to an OS level directory listing command like ls or dir.

However, I find it odd that specifying a file wildcard matches on
directories. For instance (DIRECTORY (MAKE-PATHNAME :DIRECTORY NIL :NAME
:WILD :TYPE :WILD) :DIRECTORIES T :FILES NIL) will return all directories.
I would expect this to always return NIL. I'm asking for just directories
which ha

Re: Bignum issue

R. Matthew Emerson — 2012-05-15T17:43:17


[...]

Gary created http://trac.clozure.com/ccl/ticket/968 for this.

Got it (was: Capturing output from shell command)

pekka — 2012-05-14T21:54:20

Forget it,

I managed to solve the problem with the following code:

---------------------------
(defun main (args)
   (if args
       (format t "~&Arguments were: ~a" args))
   (system "ls" "-n")
   (quit))

(defun system (cmd &optional args)
   (with-input-from-string
       (istream (with-output-to-string (ostream)
         (run-program cmd (list args) :output ostream)))
     (copy-stream istream)))

(defun copy-stream (in)
   (loop for line = (read-line in nil nil)
        while line
        do (format t "~%~a" line)))

(let ((args ccl:*unprocessed-command-line-arguments*))
   (main args))

---------------------------

-pekka-

Capturing output from shell command

pekka — 2012-05-13T21:08:58

Dear Sir,

I reused some code from manual in order to study
command line arguments and calling external programs:

------------------------
(defun main (args)
    (format t "~a" args)
    (process-arguments)
    (quit))

(defun process-arguments ()
   (with-output-to-string (stream)
     (run-program "ls" '("-n") :wait t :output t)))

(let ((args ccl:*unprocessed-command-line-arguments*))
   (main args))

------------------------

It runs on OS X Lion terminal as follows:

$ ccl64 -l op.lisp -- one two three
total 16
-rw-r--r--  1 501  20  1280 May 13 19:11 clg.lisp
-rw-r--r--  1 501  20   270 May 13 23:49 op.lisp
(one two three)
$

All fine, except when I replace ":output t" with ":output stream".
How can I get the lines the shell directed to "stream" into a list?

------------------------
(defun main (args)
    (format t "~a" args)
    (process-arguments)
    (quit))

(defun process-arguments ()
   (with-output-to-string (stream)
     (run-program "ls" '("-n") :wait t :output stream)))
     ;; How to acces

Re: Control-C or SIGINT

Andreas Thiele — 2012-05-15T12:15:22

Sorry, widening my searches to the whole web revealed

(setf ccl:*break-hook*  
  (lambda (cond hook)                               
    (declare (ignore cond hook)) 
    (format t "Cleaning up ...") 
    (ccl:quit)))

which perfectly solves my problem.

Andreas

Control-C or SIGINT

Andreas Thiele — 2012-05-15T11:45:21

Possibly a silly question:

I wrote a tiny linux console application (tool). This tool processes a file,
creates output and waits until the file further grows to present more output
(I use my program fwflt in a pipe: tail -f /var/log/xx | fwflt -c).

I'd like to simply break the program using control-c at the console prompt.
This works fine when my program waits for the file to grow - but - when I
hit control-c during an output/processing phase, everything gets messed up.
A break listener loop is created and terminal input to the repl seems to be
(non sensibly) fed from my programs terminal output.

Is there a way out of this problem? 

I used save-application with :error-handler :quit-quietly

Hints are appreciated :)

Best Regards Andreas

Bignum issue

James M. Lawrence — 2012-05-15T00:09:47

This eventually gives an assertion failure on my machine.

(defun compute ()
  (loop
     :for result := 1 :then (* result 10)
     :repeat 2000
     :finally (return result)))

(defun test ()
  (declare (optimize (debug 3) (safety 3) (speed 0)))
  (let ((expected (expt 10 2000))
        (result (compute)))
    (assert (eql expected result))))

(defun run ()
  (let ((count 0))
    (loop
       (test)
       (incf count)
       (when (zerop (mod count 1000))
         (princ ".")))))

Tested on 32-bit CCL 1.8 and from SVN; Linux Core-i7. Number of
iterations until the assert signals: 36000, 31000, 18000, 24000,
230000 (an outlier).

The value of `result' varies, though it always consists of a bunch of
nines, followed by random digits, followed by zeros. The value of
`expected' is correct.

The problem seems to be exacerbated by the presence of threads. If we
create many threads then the assertion fails immediately.

(defun test ()
  (declare (optimize (debug 3) (safety 3) (speed 0)))
  (let ((expected (expt 10

Re: Semaphore troubles

Erik Pearson — 2012-05-10T22:13:54

Ordinarily, extra time for sleeping is a good thing :)

On Thu, May 10, 2012 at 3:08 PM, Gary Byers <gb< at >clozure.com> wrote:

_______________________________________________
Openmcl-devel mailing list
Openmcl-devel< at >clozure.com
http://clozure.com/mailman/listinfo/openmcl-devel