From marko.mahnic at email.si  Tue Oct  4 06:11:40 2005
From: marko.mahnic at email.si (Marko Mahnic)
Date: Tue Jan 30 08:52:09 2007
Subject: [slang-users] slirp 1.7.5 released
In-Reply-To: <20050928172336.GA24417@svoboda.mit.edu>
References: <20050928172336.GA24417@svoboda.mit.edu>
Message-ID: <4342555C.6050907@email.si>

Michael Noble wrote:
> Hello Folks,
> 
> Version 1.7.5 of SLIRP is now available at
...
> 13.	SLIRP now installs files to its own directory within $prefix/share,
> 	instead of $prefix/share/slsh, in part to fix the autoloading bug

It still does not work for me. I am working on cygwin.
I installed with:

$ ./configure
$ make
$ make install

prefix		    = /usr/local
exec_prefix	    = ${prefix}
install_bin_dir     = ${exec_prefix}/bin
install_pkg_dir     = ${prefix}/share/slirp


$ slsh --version
slsh version 0.7.3-0
S-Lang Library Version: 2.0.4

$ slsh -v
Loading /usr/local/etc/slsh.rc
Loading /usr/local/share/slsh/autoload.sl
Loading /usr/local/share/slsh/slshrl.sl

$ slirp -v
Unable to locate autoload.sl on load path
Traceback: evalfile
/usr/local/etc/slsh.rc:43:<top-level>:Open failed

If I run slirp from /usr/local/share/slirp, it works.
If I remove /usr/local/etc/slsh.rc, I get

$ slirp -v
require is undefined
/usr/local/share/slirp/slirp.sl:31:<top-level>:Undefined Name


I guess the problem is in the directory search list:
    /usr/local/etc/slsh.rc
is found, but
    /usr/local/share/slirp/slsh.rc
isn't.

How to fix this?

Marko


From marko.mahnic at email.si  Tue Oct  4 07:04:45 2005
From: marko.mahnic at email.si (Marko Mahnic)
Date: Tue Jan 30 08:52:09 2007
Subject: [slang-users] slirp 1.7.5 released
In-Reply-To: <4342555C.6050907@email.si>
References: <20050928172336.GA24417@svoboda.mit.edu>
	<4342555C.6050907@email.si>
Message-ID: <434261CD.7020700@email.si>

Marko Mahnic wrote:
> Michael Noble wrote:
> 
>>Hello Folks,
>>
>>Version 1.7.5 of SLIRP is now available at
> 
> ...
> 
>>13.	SLIRP now installs files to its own directory within $prefix/share,
>>	instead of $prefix/share/slsh, in part to fix the autoloading bug
> 
> 
> It still does not work for me. I am working on cygwin.
> I installed with:
> 

Sory.
This was an installation problem on my side. I had slirp
1.7.4 installed and after calling make uninstall (1.7.4) it did
not uninstall properly: the file slirpsh.exe was left behind.

The problem was in the hard-coded search path list.
Default search path: /usr/local/share/slsh/local-packages:.:/cygdrive/f/build/slirp-1.7.4

After removing slirpsh.exe, all is well.

Marko


From mnoble at space.mit.edu  Tue Oct  4 08:54:27 2005
From: mnoble at space.mit.edu (Michael Noble)
Date: Tue Jan 30 08:52:09 2007
Subject: [slang-users] slirp 1.7.5 released
In-Reply-To: <434261CD.7020700@email.si>
References: <20050928172336.GA24417@svoboda.mit.edu>
	<4342555C.6050907@email.si> <434261CD.7020700@email.si>
Message-ID: <20051004125427.GA8724@svoboda.mit.edu>

> This was an installation problem on my side. I had slirp
> 1.7.4 installed and after calling make uninstall (1.7.4) it did
> not uninstall properly: the file slirpsh.exe was left behind.

Thanks, Marko.  I'm away at a conference now, but have made a
note of this and will ensure it does not persist.

-Mike


From joerg at alea.gnuu.de  Fri Oct  7 07:50:56 2005
From: joerg at alea.gnuu.de (Joerg Sommer)
Date: Tue Jan 30 08:52:09 2007
Subject: [slang-users] Bug in namespace allocation
Message-ID: <slrndkco8v.1qg.joerg@alea.gnuu.de>

Hi John,

#v+
$ cat /tmp/test.sl 
% autoload("foo", "/tmp/test1.sl");
autoload("_test->bar", "/tmp/test1.sl");

% foo();
_test->bar();
$ cat /tmp/test1.sl 
% try whether reevaluation works in SLANG 2
implements("_test");
message(current_namespace());

public define foo()
{
    message("foo");
}

define bar()
{
    message("bar");
}
$ slsh /tmp/test.sl 
Namespace _test already exists
/tmp/test1.sl:2:<top-level>:Namespace Error
#v-

I've added a debug output to implements_ns() and saw that:
ns->name = " *** internal ns <1> *** "
name = "/tmp/test1.sl"

in the error case. I think the allocation of the namespace by autoload()
is the problem.

Bye, J?rg.

BTW: Do you have a Bug Tracking System?
-- 
Die meisten Menschen wollen lieber durch Lob ruiniert
als durch Kritik gerettet werden.


From davis at space.mit.edu  Fri Oct  7 11:48:17 2005
From: davis at space.mit.edu (John E. Davis)
Date: Tue Jan 30 08:52:09 2007
Subject: [slang-users] Bug in namespace allocation
In-Reply-To: <slrndkco8v.1qg.joerg@alea.gnuu.de>
References: <slrndkco8v.1qg.joerg@alea.gnuu.de>
Message-ID: <200510071548.j97FmHFR020112@aluche.mit.edu>

Joerg Sommer <joerg@alea.gnuu.de> wrote:
>$ cat /tmp/test.sl 
>% autoload("foo", "/tmp/test1.sl");
>autoload("_test->bar", "/tmp/test1.sl");
>
>% foo();
>_test->bar();
>$ cat /tmp/test1.sl 
>% try whether reevaluation works in SLANG 2
>implements("_test");
>message(current_namespace());
>
>public define foo()
>{
>    message("foo");
>}
>
>define bar()
>{
>    message("bar");
>}
>$ slsh /tmp/test.sl 
>Namespace _test already exists
>/tmp/test1.sl:2:<top-level>:Namespace Error
>#v-
>
>I've added a debug output to implements_ns() and saw that:
>ns->name = " *** internal ns <1> *** "
>name = "/tmp/test1.sl"
>
>in the error case. I think the allocation of the namespace by autoload()
>is the problem.

I am not convinced that it is a problem.  For example, try removing
the implements line from test1.sl.  You will find it works:

   $ slsh /tmp/test.sl
   _test
   bar

As I see it, the real problem is the use of `implements`, which I
regard as unnecessary for most purposes.  For example, by omitting the
implements line, test.sl can be changed to:

   autoload("foo->bar", "/tmp/test1.sl");
   foo->bar();

to produce the output:

   foo
   bar

The upshot is that I think that it is far better for the "loader" to
specify the namespace because only the "loader" can know what
namespaces already exist.

>BTW: Do you have a Bug Tracking System?

Yes and no.  No, I do not use any sort of automated bug tracking
software.  Yes, because when a bug report comes in, I make a note of the
problem and attempt to fix it.  It is also worth pointing out systems
like debian do track bugs on all of its "upstream" packages.  For
example, look at debian's bug tracking lists for slang1 and slang2:

  <http://bugs.debian.org/cgi-bin/pkgreport.cgi?pkg=slang1-dev>
  <http://bugs.debian.org/cgi-bin/pkgreport.cgi?pkg=libslang2-dev>

--John


From joerg at alea.gnuu.de  Fri Oct  7 18:56:39 2005
From: joerg at alea.gnuu.de (=?iso-8859-1?Q?J=F6rg?= Sommer)
Date: Tue Jan 30 08:52:09 2007
Subject: [slang-users] Bug in namespace allocation
In-Reply-To: <200510071548.j97FmHFR020112@aluche.mit.edu>
References: <slrndkco8v.1qg.joerg@alea.gnuu.de>
	<200510071548.j97FmHFR020112@aluche.mit.edu>
Message-ID: <20051007225639.GA3213@alea.gnuu.de>

Hello John,

John E. Davis schrieb am Fri 07. Oct, 11:48 (-0400):
> I am not convinced that it is a problem.  For example, try removing
> the implements line from test1.sl.  You will find it works:
> 
>    $ slsh /tmp/test.sl
>    _test
>    bar

But this leads to very strange errors:

#v+
$ cat /tmp/test.sl 
autoload("foo", "/tmp/test1.sl");
autoload("_test->bar", "/tmp/test1.sl");

define bar()
{
    message("bar not in test1.sl");
}
bar();

foo();
bar();
_test->bar();

message("");
foreach (_get_namespaces())
{
    loop ( _apropos( (), "bar", 0xF) ) {
        "->"; exch();
        () + () + ();
        message( () );
    }
}
$ cat /tmp/test1.sl 
% try whether reevaluation works in SLANG 2
% implements("_test");
message(current_namespace());

public define foo()
{
    message("foo");
}

define bar()
{
    message("bar");
}

$ slsh /tmp/test.sl 
bar not in test1.sl

foo
bar
_test
bar

_test->bar
Global->bar
#v-

Global->bar from test.sl gets overwritten because no namespace is given
to the autoload() for foo. So you can't refer to the public functions in
a unit with autoload(), otherwise other functions in the Global namespace
may become overwritten.

Regards, J?rg.
-- 
Perfection is reached, not when there is no longer anything that can be
added, but when there is no longer anything to take away.
                                        (Antoine de Saint-Exupery)
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 477 bytes
Desc: not available
Url : http://mailman.jtan.com/pipermail/slang-users-l/attachments/20051008/f9dbc0c9/attachment.bin
From davis at space.mit.edu  Sat Oct  8 16:44:21 2005
From: davis at space.mit.edu (John E. Davis)
Date: Tue Jan 30 08:52:09 2007
Subject: [slang-users] Bug in namespace allocation
In-Reply-To: <20051007225639.GA3213@alea.gnuu.de>
References: <slrndkco8v.1qg.joerg@alea.gnuu.de>
	<200510071548.j97FmHFR020112@aluche.mit.edu>
	<20051007225639.GA3213@alea.gnuu.de>
Message-ID: <200510082044.j98KiLVD011466@aluche.mit.edu>

J?rg Sommer <joerg@alea.gnuu.de> wrote:
>But this leads to very strange errors:

I would not call this behavior "strange" nor an "error".  It works as
it is documented to work.  What you seem to be looking for is a way to
autoload a public function from a file evaluated in a specified
namespace.  Currently there exists no simple mechanism to perform such
a task.  However, it is rather easy to create a function that achieves
this result.  Let's call it "autoload_public".  Then in test.sl
replace

   autoload("foo", "/tmp/test1.sl");

with

   autoload_public ("foo", "/tmp/test1.sl", "_test");

This statement says to autoload "foo" from "/tmp/test.sl" evaluated in
the "_test" namespace.

Here is an implementation of such a function:

   define autoload_public (fun, file, ns)
   {
      if (is_defined (fun))
	return;

      eval ("public define $fun();"$
	    + "public define $fun(){"$
	    + "variable args=__pop_args(_NARGS);"
	    + "()=evalfile(\"$file\",\"$ns\");"$
	    + "return $fun(__push_args(args));}"$);
   }

--John


From davis at space.mit.edu  Mon Oct 17 13:17:20 2005
From: davis at space.mit.edu (John E. Davis)
Date: Tue Jan 30 08:52:09 2007
Subject: [slang-users] slang 2.0.5 released
Message-ID: <200510171717.j9HHHK1Z028923@aluche.mit.edu>

Version 2.0.5 is available.  Although it is primarily a bug-fix
release, it contains some performance tweaks and some new intrinsic
functions.

See <http://www.jedsoft.org/slang/download.html> for download options.

Here are the relevant MD5 sums:

   377ccd8232bd3c240f51a4a41ae0a426  2.0.4__2.0.5.diff.bz2
   8ae18681473f3c0c5771245b2e380df9  2.0.4__2.0.5.diff.gz
   8b6afa085f76b1be29825f0c470b6cad  slang-2.0.5.tar.bz2
   a80801b10453ca317e8a9ae8e270163e  slang-2.0.5.tar.gz

Changes since 2.0.4
1. slsh/slsh.c: Error messages were not being printed if an error
   occured during an atexit hook.
2. src/slstrops.c: Use _pSLstring_bytelen instead of strlen to get the
   number of bytes.
3. src/slstring.c: Several tweaks were made for faster string handling.
4. src/slparse.c: A negative integer literal raised to a power was not
   being properly parsed.
5. doc/tm/cslang.tm: Documentation updated to reflect the changed
   SLang_pop_double API, as well as other C API documentation
   tweaks (Doug Burke).
6. doc/tm/rtl/type.tm: Documentation for "integer" changed to indicate
   that integer throws a SyntaxError exception if the string does not
   have a valid syntax. (Joshua Kempner jkempner at bowdoin, edu)
7. src/slmath.c: Added nint (round to the nearest integer) intrinsic.
8. src/slarray.c: NULL arguments to array_map were not getting
   converted to arrays of NULL.
9. src/slang.h: Added SLclass_set_length_function for setting the
   length method of application-defined classes.
10. src/slang.ver: A versioning script is used to restrict the
    visibility of the _pSL* symbols.  (Alastair McKinstry
    mckinstry at computer, org)
11. slsh/slsh.c: Ignore SIGPIPE.
12. src/slutf8.c: If SLutf8_decode is passed an inproperly encoded
    UTF-8 sequence, then set the target of the SLwchar_Type pointer to
    the ascii value of the byte, in addition to returning NULL.
13.  src/slposdir.c: Added "access" intrinsic.
14. src/slsmg.c: Fixed a buffer-overflow in the SLsmg_write_chars
    function.
15. src/slsig.c: A missing return statement was causing sigsuspend to
    fail when called with 0 arguments.
16. src/slarray.c: Added wherefirst and wherelast intrinsics.
17. src/slang.c: Inline attribute removed from some functions that
    were failing to be inlined by the compiler.
19. src/slstdio.c: SLang_get_fileptr added to permit an application to
    get the FILE* object associated with a SLANG_FILE_PTR_TYPE.
20. src/slang.c: It was failing to compile when
    SLANG_HAS_DEBUGGER_SUPPORT was set to 0.
21. src/sltime.c: strftime intrinsic added.


From giva at bgnett.no  Mon Oct 17 16:11:09 2005
From: giva at bgnett.no (Gisle Vanem)
Date: Tue Jan 30 08:52:09 2007
Subject: [slang-users] [Patch] slvideo.c
Message-ID: <094701c5d356$e9e45270$0600000a@broadpark.no>

A little patch for djgpp is needed in slvideo.c. I haven't
checked the other DOS compilers. Is OpenWatcom on
DOS still possible?

--- orig/slvideo.c 2005-10-17 05:00:18.000000000 +0200
+++ ./slvideo.c 2005-10-17 21:22:08.000000000 +0200
@@ -818,18 +818,16 @@
 static void
 write_attributes (SLsmg_Char_Type *src, unsigned int count)
 {
-   register unsigned short pair;
    unsigned int n;

    /* write into a character/attribute pair */
    n = Cursor_Col;
    while (count)
      {
-       pair = SLSMG_CHAR_TO_USHORT(*src);/* character/color pair */
-       src++;
-       SLtt_reverse_video (pair >> 8); /* color change */
-       ScreenPutChar ((int)pair & 0xFF, Attribute_Byte, n, Cursor_Row);
+        SLtt_reverse_video (src->color); /* color change */
+        ScreenPutChar (src->wchars[0] & 0xFF, Attribute_Byte, n, Cursor_Row);
        n++;
+        src++;
        count--;
      }
 }

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

--gv


From perez.l at free.fr  Tue Oct 18 15:35:29 2005
From: perez.l at free.fr (Laurent Perez)
Date: Tue Jan 30 08:52:09 2007
Subject: [slang-users] Import Error with PNG module
Message-ID: <20051018193529.GB2495@lincoln.org>

Hi,

I got this error when importing png module :

slsh> require ("png");
Error linking to /home/laurent//lib/slang/v2/modules/png-module.so:
/usr/lib/libpng.so.3: undefined symbol: deflate
/home/laurent/share/slsh/png.Sl:2:<top-level>:Import Error

As far as I know, deflate is defined in the zlib library used by png. I
don't know if it is a system error (Slackware 10.1) but changing the
Makefile to force linking with zlib ( PNG_LIB =  -lpng -lz ) made a
working png module.

ldd png-module.so                       
        libpng.so.3 => /usr/lib/libpng.so.3 (0x4001a000)
        libz.so.1 => /usr/lib/libz.so.1 (0x4004c000)
        libm.so.6 => /lib/libm.so.6 (0x4005d000)
        libc.so.6 => /lib/libc.so.6 (0x40080000)
        /lib/ld-linux.so.2 (0x80000000)


Laurent.


From davis at space.mit.edu  Tue Oct 18 17:14:06 2005
From: davis at space.mit.edu (John E. Davis)
Date: Tue Jan 30 08:52:09 2007
Subject: [slang-users] Import Error with PNG module
In-Reply-To: <20051018193529.GB2495@lincoln.org>
References: <20051018193529.GB2495@lincoln.org>
Message-ID: <200510182114.j9ILE6OQ022043@aluche.mit.edu>

Laurent Perez <perez.l@free.fr> wrote:
>As far as I know, deflate is defined in the zlib library used by png. I
>don't know if it is a system error (Slackware 10.1) but changing the
>Makefile to force linking with zlib ( PNG_LIB =  -lpng -lz ) made a
>working png module.

I will add -lz to the Makefile for the next release.  FWIW, debian
sarge does not have this problem because of the way zlib was built and
installed:

# ldd /usr/lib/libpng.so.2
	libz.so.1 => /usr/lib/libz.so.1 (0x4002c000)
	libm.so.6 => /lib/tls/libm.so.6 (0x4003e000)
	libc.so.6 => /lib/tls/libc.so.6 (0x40061000)
	/lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x80000000)

Thanks,
--John


From perez.l at free.fr  Thu Oct 20 08:34:18 2005
From: perez.l at free.fr (Laurent Perez)
Date: Tue Jan 30 08:52:09 2007
Subject: [slang-users] slgtk with slang-2
Message-ID: <20051020123418.GA18316@lincoln.org>

Hi,

Did anybody succeed in compiling slgtk with slang-2 ? I tried using slirp
1.7.5 and slirp included with slgtk-0.5.14 but it failed with this error :

Generating Gtk wrapper code: make[1]: *** [.gencode] Error 30

I don't ever know if slirp can be used with slang-2.

Thanks for your help.


Laurent.


From mnoble at space.mit.edu  Thu Oct 20 09:42:07 2005
From: mnoble at space.mit.edu (Michael Noble)
Date: Tue Jan 30 08:52:09 2007
Subject: [slang-users] slgtk with slang-2
In-Reply-To: <20051020123418.GA18316@lincoln.org>
References: <20051020123418.GA18316@lincoln.org>
Message-ID: <20051020134207.GA22968@svoboda.mit.edu>

> Did anybody succeed in compiling slgtk with slang-2 ? I tried using slirp
> 1.7.5 and slirp included with slgtk-0.5.14 but it failed with this error :

Yes, I build it on a regular basis against SLang2.  In fact, at

	ftp://space.mit.edu/pub/mnoble/slgtk/binaries/

you'll find that
  
	slgtk-0.5.14-i686-slang2.tar.gz

is a set of binaries for such (built on Debian Woody, with GCC 2.95).

> Generating Gtk wrapper code: make[1]: *** [.gencode] Error 30

Can you tell me what platform you're on?  What compiler you're using?
And what version of Gtk?  Also, what does

	tail -20 ignoredfuncs.txt

(in the src subdirectory) show?


Thanks,
Mike


From robertlipe at usa.net  Sun Oct 23 18:09:50 2005
From: robertlipe at usa.net (Robert Lipe)
Date: Tue Jan 30 08:52:09 2007
Subject: [slang-users] Advice on "complex" data type integration
Message-ID: <20051023220950.GA23481@rjloud.sco.com>

Hello,

I've been prototyping integration of S-Lang into GPSBabel.  I'm
impressed with the power of the presented language; it's an nice piece
of work.  Unfortunately, I'm getting all tripped up in integrating it
and am looking for some advice beyond the "C Programmer's Guide".

I have several internal data types.  They aren't horribly complicated,
but they're not trivial.  Some are dynamically generated (a waypoint
from Delorme software may have additional traits that are different
from the additional traits of a waypoint from certain XML formats).
Internally, there is extensive use of linked lists and structures hooked
together with other linked lists.

Since my "real" structures have a number of enumerated types, structures
containing structures, and other things that make themn a clunky fit to
MAKE_[IC]STRUCT_FIELD and friends, I'm kind of stumped on how to map
these on the way across the Slang border.

I'd considered making a "visible to slang" structure and exposing an
interator to the language that did a copy from the "real" C struct to
the the type I expose to make_[ci]struct_table, but that seems pretty
klunky and performance would have to be a killer.

        variable i = 0;
        for (i = gb_waypt_start(); i; i = gb_waypt_next()) {
                messsage(string(wpt.shortname));
        }

So if I plumb the "gb_" brothers above to call SLadd_istruct_table on
the struct and I copy every single member on the way in and out, it's
workable but really kind of gross.

The internal structure has things like enumerations that should probably
be exposed to slang as strings.  I have things like a time that's a
time_t plus an int for hundredths of a second, so I'd need to be able to
"shim" the timestamps on the way in and out.

So given my data types at
  http://cvs.sourceforge.net/viewcvs.py/gpsbabel/gpsbabel/defs.h?view=markup
for struct waypoint,  I'd like to be able to write Slang fragments like:

        wpt.icon = "Foo";

to set wpt_flags.icon_descr_is_dynamic, call my free routine if it was
set before, and then morph your SLang_create_slstring'ed object into one
created by my xmalloc object.

        blah = string(wpt.gc_container);

would have to read the gc_data structure in the waypoint structure, see
that gc_container is a bitfield, then call an internal function to get
the string representation of it.

        routewpt.format("an1", "line_color") = rgb

I haven't come to grips with the syntax there, but this would have to
walk through the extra data list until it found the 'an1' items attached
to this routewpt, then find the 'line_color' member in THAT struct and
generate a store to it.


Additionally, I have my own internal memory management and requiring
every string field be set and cleared via SLang_create_slstring and
SLang_free_slstring is a problem.  Is there a a way I can chain into
those two in my allocator?


I guess I could do the whole conversion at the very entry point and
exit points of the interpreter, copying the entire list to Slang-compat
types and then "undoing" it on the way back, but it just seems like I'm
getting further off balance here.

What's the most sensible way to map these things to Slang?

I've looked through several of the examples and I'm not seeing anything
at a glance that maps very well to my case that I can use for guidance.
I'd love to be able to get the power of SLang, but I'm not going to
replace my data structures and all accesses to them to get it.  Given
the elgance of the interfaces, I'm supposing there's a way to get there
from here, so I'd appreciate any pointers that can be offered.  Do you
have a good example on file of mapping things beyond simple C structs
into Slang?

Thanx in advance,
RJL



From davis at space.mit.edu  Sun Oct 23 18:40:14 2005
From: davis at space.mit.edu (John E. Davis)
Date: Tue Jan 30 08:52:09 2007
Subject: [slang-users] Advice on "complex" data type integration
In-Reply-To: <20051023220950.GA23481@rjloud.sco.com>
References: <20051023220950.GA23481@rjloud.sco.com>
Message-ID: <200510232240.j9NMeEJg027391@aluche.mit.edu>

Robert Lipe <robertlipe@usa.net> wrote:
>So given my data types at
>  http://cvs.sourceforge.net/viewcvs.py/gpsbabel/gpsbabel/defs.h?view=markup
>for struct waypoint,  I'd like to be able to write Slang fragments like:
[...]
>from here, so I'd appreciate any pointers that can be offered.  Do you
>have a good example on file of mapping things beyond simple C structs
>into Slang?

Have you considered using slirp <http://space.mit.edu/~mnoble/slirp/>
to wrap the functions and data structures in this file?  It should
produce something that is more or less a straightforward mapping of
your interface, and handle all the memory management for you.
Mike Noble is the slirp expert and can probably provide additional
details.

Thanks,
--John


From robertlipe at usa.net  Sun Oct 23 21:27:29 2005
From: robertlipe at usa.net (Robert Lipe)
Date: Tue Jan 30 08:52:09 2007
Subject: [slang-users] Advice on "complex" data type integration
In-Reply-To: <200510232240.j9NMeEJg027391@aluche.mit.edu>
References: <20051023220950.GA23481@rjloud.sco.com>
	<200510232240.j9NMeEJg027391@aluche.mit.edu>
Message-ID: <20051024012729.GR3824@rjloud.sco.com>

> >  http://cvs.sourceforge.net/viewcvs.py/gpsbabel/gpsbabel/defs.h?view=markup
> >for struct waypoint,  I'd like to be able to write Slang fragments like:
> [...]
> 
> Have you considered using slirp <http://space.mit.edu/~mnoble/slirp/>
> to wrap the functions and data structures in this file?  It should

Thanx for the tip.  I'm still digesting it to see how that would
improve things.  

I was hoping to not actually expose the bits and the bytes of the
internal data structures out to the Slang side.  I have comparitively
little _code_ I want to expose, I'm mostly interested in data.

Another approach (which doesn't thrill me) that came to me after
struggling with this for a while was to just bring out acessors with
arguments of strings.

	linkylink = gb_get_wpt(wpt, "url");
	gb_set_wpt(wpt, "url", "http://" + linkylink);


...and have the C code look up the data thingy being accessed and Do
The Right Thing.  That's not real nice from either the linguistic or
maintenance view on either the SLang side or the C side.

I'll continue trying to get my head around slirp.  Thanx.

RJL


From davis at space.mit.edu  Sun Oct 23 22:16:43 2005
From: davis at space.mit.edu (John E. Davis)
Date: Tue Jan 30 08:52:09 2007
Subject: [slang-users] Advice on "complex" data type integration
In-Reply-To: <20051024012729.GR3824@rjloud.sco.com>
References: <20051023220950.GA23481@rjloud.sco.com>
	<200510232240.j9NMeEJg027391@aluche.mit.edu>
	<20051024012729.GR3824@rjloud.sco.com>
Message-ID: <200510240216.j9O2Ghf5030703@aluche.mit.edu>

Robert Lipe <robertlipe@usa.net> wrote:
>Another approach (which doesn't thrill me) that came to me after
>struggling with this for a while was to just bring out acessors with
>arguments of strings.
>
>	linkylink = gb_get_wpt(wpt, "url");
>	gb_set_wpt(wpt, "url", "http://" + linkylink);

It is fairly easy to construct an interface that permits the above to
be written:

   wpt = gb_new_wpt ();
   wpt.url = "http://" + wpt.url;

and have the wpt object to be automatically freed when it goes out of
scope or reassigned.  To do the above, you would use the sput and sget
class methods.  Is this something more akin to what you are looking
for?

Thanks,
--John


From robertlipe at usa.net  Mon Oct 24 08:52:12 2005
From: robertlipe at usa.net (Robert Lipe)
Date: Tue Jan 30 08:52:09 2007
Subject: [slang-users] Advice on "complex" data type integration
In-Reply-To: <200510240216.j9O2Ghf5030703@aluche.mit.edu>
References: <20051023220950.GA23481@rjloud.sco.com>
	<200510232240.j9NMeEJg027391@aluche.mit.edu>
	<20051024012729.GR3824@rjloud.sco.com>
	<200510240216.j9O2Ghf5030703@aluche.mit.edu>
Message-ID: <20051024125212.GV3824@rjloud.sco.com>

>    wpt = gb_new_wpt ();
>    wpt.url = "http://" + wpt.url;
> 
> and have the wpt object to be automatically freed when it goes out of
> scope or reassigned.  To do the above, you would use the sput and sget
> class methods.  Is this something more akin to what you are looking

If I could "hook" the read and the writes of the structure members (and,
presumably, the creation and destruction of wpt) I think I could work
with that.

I see references to cl_s(ge|pu)t and SLclass_set_s(ge|pu)t_function in 
slang.h but I don't see them in http://www.s-lang.org/doc/html/cref.html#toc5 .

I recall being intrigued by the class functions in Chapter 5, but they
punted to http://www.s-lang.org/doc/html/cslang.html#toc5 for more
information, yet I couldn't find any and none of the S-Lang-using
projects I checked used them.  So I assumed these were internals that
just leaked out.

If you have doc or examples on these, and think this is a pond worth
fishing in, I'd like to check them out.

Thanx for the response, John.

RJL


From davis at space.mit.edu  Mon Oct 24 13:05:22 2005
From: davis at space.mit.edu (John E. Davis)
Date: Tue Jan 30 08:52:09 2007
Subject: [slang-users] Advice on "complex" data type integration
In-Reply-To: <20051024125212.GV3824@rjloud.sco.com>
References: <20051023220950.GA23481@rjloud.sco.com>
	<200510232240.j9NMeEJg027391@aluche.mit.edu>
	<20051024012729.GR3824@rjloud.sco.com>
	<200510240216.j9O2Ghf5030703@aluche.mit.edu>
	<20051024125212.GV3824@rjloud.sco.com>
Message-ID: <200510241705.j9OH5MQt009853@aluche.mit.edu>

Robert Lipe <robertlipe@usa.net> wrote:
>If you have doc or examples on these, and think this is a pond worth
>fishing in, I'd like to check them out.

JED <http://www.jedsoft.org/jed/> makes use of the sget functions
for its "Mark_Type" objects representing buffer positions.  It also
defines a number of binary operators for these objects allowing
comparisons of buffer positions, e.g.,

   if (m1 > m2)
     (m1, m2) = (m2, m1);   % swap the marks

The marks are represented as SLANG_CLASS_TYPE_MMT objects so that the
memory associated with them is managed by the interpreter.  This means
that such objects will automatically be freed (by calling the appropriate
destroy method) when nolonger used.

The code may be found in paste.c in the jed distribution.

Perhaps this coming weekend I will add a more complete example to the
demo directory.  But I think the code in paste.c should be enough to
get you started.

--John