headers
Eric Marsden | 3 Nov 1999 13:02
Picon

massive heap consumption with ,load-package environments

Hi,

This is probably a misunderstanding of mine rather than a bug. I am
trying to access the to-cps package defined in
ps-compiler/prescheme/package-defs.scm. I load the interfaces and
package definitions from various areas to define the different things
which are required, and say

   ,load-package environments

after which scheme48 starts growing the heap apparently indefinitely
(even when started with a heap of 400MB). The last messages are

   [number-i/o scheme/rts/numio.scm]
   [fluids scheme/rts/fluid.scm
   Analyzing... 
   Calls will be compiled in line: (empty-dynamic-env make-thread
   thread-dynamic-point thread-dynamic-env make-fluid
   fluid-top-level-value set-fluid-top-level-value!)
   Scheme48 heap overflow

Here are the things I loaded:

   scheme/interfaces.scm
   scheme/more-interfaces.scm
   scheme/alt-packages.scm
   scheme/comp-packages.scm
   scheme/initial-packages.scm
   scheme/link-packages.scm
   scheme/low-packages.scm
(Continue reading)

Permalink | Reply | 3 comments |
headers
Richard Kelsey | 3 Nov 1999 13:58
Picon

Re: massive heap consumption with ,load-package environments

   From: Eric Marsden <emarsden <at> mail.dotcom.fr>
   Date: 03 Nov 1999 13:02:15 +0100

   This is probably a misunderstanding of mine rather than a bug. I am
   trying to access the to-cps package defined in
   ps-compiler/prescheme/package-defs.scm. I load the interfaces and
   package definitions from various areas to define the different things
   which are required, and say

      ,load-package environments

   after which scheme48 starts growing the heap apparently indefinitely
   (even when started with a heap of 400MB).

What version of Scheme 48 are you using?

                                       -Richard Kelsey
Permalink | Reply | 1 comment |
headers
Eric Marsden | 3 Nov 1999 14:26
Picon

Re: massive heap consumption with ,load-package environments

>>>>> "rk" == Richard Kelsey <kelsey <at> research.nj.nec.com> writes:

  rk> What version of Scheme 48 are you using?

sorry, v0.53 (on linuxppc and Solaris, though I don't suppose it
matters). 

--

-- 
Eric Marsden
Permalink | Reply | 0 comments |
headers
Richard Kelsey | 3 Nov 1999 16:01
Picon

Re: massive heap consumption with ,load-package environments


   From: Eric Marsden <emarsden <at> mail.dotcom.fr>
   Date: 03 Nov 1999 13:02:15 +0100
   Lines: 44

   This is probably a misunderstanding of mine rather than a bug. I am
   trying to access the to-cps package defined in
   ps-compiler/prescheme/package-defs.scm. I load the interfaces and
   package definitions from various areas to define the different things
   which are required, and say

      ,load-package environments

   after which scheme48 starts growing the heap apparently indefinitely
   (even when started with a heap of 400MB).

You are undoubtably stuck in an infinite recursion.

  The last messages are

      [number-i/o scheme/rts/numio.scm]
      [fluids scheme/rts/fluid.scm
      Analyzing... 
      Calls will be compiled in line: (empty-dynamic-env make-thread
      thread-dynamic-point thread-dynamic-env make-fluid
      fluid-top-level-value set-fluid-top-level-value!)
      Scheme48 heap overflow

My guess is that the system is confused because you are reloading
the fluids package.  The fluid code uses a VM register to hold
(Continue reading)

Permalink | Reply | 0 comments |
headers
Mike Wilson | 3 Nov 1999 20:29
Picon
Favicon

'-- breaks


Scheme 48 0.53

Symbols starting with two dashes break.

  Welcome to Scheme 48 0.53 (made by cmwilson on Sat Oct 23 19:50:33 EDT 1999).
  Copyright (c) 1993-1999 by Richard Kelsey and Jonathan Rees.
  Please report bugs to scheme-48-bugs <at> martigny.ai.mit.edu.
  Type ,? (comma question-mark) for help.
  > '--

  Error: unsupported number syntax
  	 "--"
  	 #{Input-port #{Input-channel "standard input"}}
  >

Mike

--

-- 
Mike Wilson    cmwilson <at> ntrnet.net    Debian GNU/Linux!
senseitachi wa boku wo  fuan ni suru kedo
sore hodo taisetsuna kotoba wa nakatta
--The Blue Hearts [shounen no shi]
Permalink | Reply | 2 comments |
headers
Richard Kelsey | 3 Nov 1999 20:57
Picon

Re: '-- breaks

   Date: Wed, 3 Nov 1999 14:29:31 -0500
   From: Mike Wilson <cmwilson <at> sarg.Ryerson.CA>

   Scheme 48 0.53

   Symbols starting with two dashes break.

     Welcome to Scheme 48 0.53 (made by cmwilson on Sat Oct 23 19:50:33 EDT 1999).
     Copyright (c) 1993-1999 by Richard Kelsey and Jonathan Rees.
     Please report bugs to scheme-48-bugs <at> martigny.ai.mit.edu.
     Type ,? (comma question-mark) for help.
     > '--

     Error: unsupported number syntax
	    "--"
	    #{Input-port #{Input-channel "standard input"}}
     >

According to R{345}RS `--' is not an identifier.  Scheme 48 only
accepts a few nonstandard identifiers (1+ and -1+ for S&ICP and
-> for Jonathan's thesis).  This is great for producing portable
code but not so hot for running code written using implementations
that have a more liberal policy.  Perhaps we should have an optional
reader that is more forgiving.  We could even allow the use of
square brackets in place of parentheses.

                                      -Richard Kelsey
Permalink | Reply | 1 comment |
headers
Mike Wilson | 3 Nov 1999 21:13
Picon
Favicon

Re: '-- breaks


) According to R{345}RS `--' is not an identifier.

Oops.  You're right.  Sorry about that.

I was trying to use Mark Johnson's LALR parser.
It uses '--> .

) We could even allow the use of
) square brackets in place of parentheses.

Hmm, well ...

Mike

--

-- 
Mike Wilson    cmwilson <at> ntrnet.net    Debian GNU/Linux!
senseitachi wa boku wo  fuan ni suru kedo
sore hodo taisetsuna kotoba wa nakatta
--The Blue Hearts [shounen no shi]
Permalink | Reply | 0 comments |

Gmane