MathGroup Archive 1997

[Date Index] [Thread Index] [Author Index]

Search the Archive

RE:FastBinaryFiles and NT4.0 REPOST

  • To: mathgroup at smc.vnet.net
  • Subject: [mg6920] RE:[mg6804] FastBinaryFiles and NT4.0 REPOST
  • From: dougb at xti.com
  • Date: Fri, 25 Apr 1997 14:00:47 -0400 (EDT)
  • Organization: XonTech
  • Sender: owner-wri-mathgroup at wolfram.com

(Also Tech Support [TS 8473] --Re:FastBinaryFiles and NT 4.0)


I previously posted: 

>Hi
>Has anybody been able to successfully load FastBinaryFiles on
>an NT 4.0 system? I've tried using Bruce Miller's recommended
>file re-names and snagging the mlink32.dll from 2.2, but I get
>an Error Message from WinNT stating that mlink32.dll is a 
>"Bad Image", then the message suggests I check the distribution 
>disk. (Was mlink32.dll compiled for Win32s ?)

>I've also been un-successful in any attempt to run a re-compiled 
>version of FastBinaryFiles.

>I get the following messages when I try my Install[""]

>MessageName::messg : OpenReadBinary::usage cannot be set to 
>OpenReadBinary[] opens the named file for reading. I ... 
>enameConverion for more information. filename, it must be set 
>to a string.

>(I get the same message with the interupting  ...  in different
>places for OpenAppendBinary and OpenWriteBinary then I get a
>General::stop followed by the following message)

>ToExpression::sntx : Syntax error in or before "    ...           ^".

>Unfortuneately, Bruce is on vacation so I haven't heard back
>from Wolfram yet.. Has anybody else been able to get 
>FastBinaryFiles to work????

>Any help would be much appreciated.
>Doug

>dougb at xti.com

Here is an update:

The binary.tm file can be compiled if the C code (that is commented as
such in the binary.tm file) is cut out and placed in a separate
binary.c file. Then run mprep on the remaining binary.tm file placing
the mprepped code into binarytm.c.

  mprep binary.tm -o binarytm.c

I then compile the code using Borland C++ 5.0 IDE (using the
description on how to compile MathLink files for Windows systems as
described in the HELP section)

I've been successful in determining the source of the  "
MessageName::messg : OpenReadBinary::usage " error messages.. If all
the usage statements are deleted from the binary.tm file prior to
running it through mprep... these messages no longer occur.

Which leaves the remaining error message -> ToExpression::sntx :
Syntax error in or before "    ...           ^".

In the binary.tm file there are the 2 following Evaluates...

:Evaluate:
 resolveNewFilename[s_String] := 
    Module[{first = First[Characters[s]], colonpos, drive, os},

 **CODE** 

   ]

:Evaluate:
 resolveExistingFilename[s_String] := 
    Module[{first = First[Characters[s]], drive, os, fileinfo, 

 **CODE** 

   ]

which after mprep has been run -> get placed inside of if(_res)
statements .. inside of the binarytm.c file

If BOTH of these if(_res) statements are commented out of the
binarytm.c code..

The resulting executable links to Mathematica no problem...

   In[1] = Install["myfastbinary", LinkMode->Launch,   
   LinkProtocol->"filemap"]

   Out[1] = LinkObject[myfastbinary,2,2]

However, there is still one tincy wincy little problem.. The
OpenWriteBinary (etc.) routines require these functions to create
their links to the output files...

Anybody have any ideas?

Doug



  • Prev by Date: Laplacian 2-D polar form
  • Next by Date: Re: Boundary Value Problem
  • Previous by thread: Re: Laplacian 2-D polar form
  • Next by thread: Re: Fingerpainting Linear Inequalities