Services & Resources / Wolfram Forums
-----
 /
MathGroup Archive
2001
*January
*February
*March
*April
*May
*June
*July
*August
*September
*October
*November
*December
*Archive Index
*Ask about this page
*Print this page
*Give us feedback
*Sign up for the Wolfram Insider

MathGroup Archive 2001

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

Search the Archive

Re: No joy using gcc 2.95.3 + MathLink v3r9 + Windows 2000

  • To: mathgroup at smc.vnet.net
  • Subject: [mg31625] Re: [mg31600] No joy using gcc 2.95.3 + MathLink v3r9 + Windows 2000
  • From: "P.J. Hinton" <paulh at wolfram.com>
  • Date: Sun, 18 Nov 2001 06:29:11 -0500 (EST)
  • Organization: "Wolfram Research, Inc."
  • Sender: owner-wri-mathgroup at wolfram.com

On Fri, 16 Nov 2001, Matthew D. Langston wrote:

> I'm trying to recompile, using gcc 2.95.3, a library of functions
> written in C/C++ that I have already successfully exported to
> Mathematica (using MathLink) with cl.exe version 6 (from Microsoft
> Visual Studio 6.0) and cl.exe version 7 (from Microsoft Visual
> Studio.NET).
>
> I'm running Windows 2000 SP2, Mathematica 4.1, and using mldev32 from
> the MathLink Developer's Kit v3r9.  The version of gcc I am using is
> 2.95.3-5 from the Cygwin 1.3.5 release (which is the latest version as I
> write this).
>
> I'm linking against the ml32i2m.lib import library (the version for
> Microsoft Visual Studio 6.0).
>
> I'm passing "-mwindows" to gcc at both compile time and link time, and
> passing "-e _mainCRTStartup" to gcc at link time.  This *should* give me
> an official, respectable Win32 application.  And indeed, these options
> do work in creating working Win32 applications.
>
> My library of C/C++ functions compiles and links fine with all three
> compilers (i.e. gcc 2.95.3, cl.exe version 6, and cl.exe version 7).  My
> functions work just fine (after Install[]ing them in Mathematica ) using
> the two Microsoft compilers, but the gcc version "appears" to hang
> Mathematica.
>
> When I run Install["Minuitgcc`"] (to load the gcc version of my MathLink
> exported functions) from within the Mathematica Front End, I get a
> dialog box with the title "MathLink" that has a "Create link:" label, a
> text entry box and an "OK" and "Cancel" button.
>
> If I click the "OK" button (without entering anything in the text box)
> then I get another dialog box with the title "MathLink" that has a "Link
> created on: 628" label and an "OK" button.
>
> My MathLink program appears in the Windows Task Manager as task
> "Minuitgcc(628)" with a status of "Running".  However, Mathematica just
> hangs on the cell that executed the Install["Minuitgcc`"] command.  The
> title of the Front End Mathematica window is "Running...m2c.nb" (my
> notebooks name is m2c.nb).
>
> The exact same code is being used by all three compilers (i.e. my source
> code of functions, the same mathlink.h file, the same output file from
> mprep, the same ml32i2m.lib import library, etc.).  The only difference
> is the compiler/linker.
>
> Why do the two Microsoft compilers create working MathLink programs,
> while gcc doesn't?

What happens if you try the following steps:

1) Launch the MathLink program Minuitgcc from the command line with the
flag "-linkcreate":

        cd <path to Minuitgcc>
        .\Minuitgcc -linkcreate

2) Note the name of the link displayed by the resulting dialog box.

3) From a Mathematica kernel, invoke the following functions:

       lnk = LinkConnect[<name of the link from (2), written as a string>]

	Install[lnk]

If this works, then the problem may have something to do with the
execution environment of Cygwin itself.  Install[] on a filename should be
the equivalent of creating a link with LinkLaunch[] and then passing the
resulting link to Install[].

-- 
P.J. Hinton
User Interface Programmer                         paulh at wolfram.com
Wolfram Research, Inc.
Disclaimer: Opinions expressed herein are those of the author alone.



  • Prev by Date: Re: No joy using gcc 2.95.3 + MathLink v3r9 + Windows 2000 SP2
  • Next by Date: Re: Re: Zero does not equal zero et al.
  • Previous by thread: RE: Re: Log plots: plot points *linearly* equidistant
  • Next by thread: problem entering text