Filter
Top Products
Chapter 4 Windows 3.1 Compiler/Linker Issues
LabWindows/CVI Programmer Reference Manual 4-2
©
National Instruments Corporation
keywords such as far, near, or huge in the declaration of functions to be called from
LabWindows/CVI source code.
• If your Watcom-compiled module performs floating point operations, you must use
Watcom Version 9.5 or later.
• Use the following options when you compile with Watcom IDE:
– Set the Project Target Environment to 32-bit Windows 3.x, and set the Image Type
to Library [.lib].
– Turn on the Disable Stack Depth Checking [-s] option.
– Turn on the Change Char Default to Signed [-j] option.
–Add
-zw -d_NI_mswin16_ to the Other Options.
– Turn on the Generate as Needed [-of] option for Stack Frames.
– Turn on the No Debugging Information option.
– Turn on the In-line with Coprocessor [fpi87] option for Floating Point Model.
– Turn on the Compiler default option for the Memory Model.
– Turn on the 80486 Stack-Based Calling [-4s] option for the Target Processor.
• Use the following compiler flags when using
wcc386 or wcc386p:
–
-zw -s -4s -j -fpi87 -d0 -of -d_NI_mswin16_
– You can use optimization flags in addition to the f, and you can use other flags, such
as
-wn, which do not affect the generation of object code.
Using 32-Bit Borland or Symantec Compiled Modules
under Windows 3.1
In this section, CVI refers to both LabWindows/CVI and Watcom modules, while Borland
applies to both Borland and Symantec modules.
The following restrictions apply to Borland object modules:
• Borland packs bit fields in structures differently than CVI, so you cannot share structures
with bit fields between Borland and CVI.
• Borland returns structures, floats, and doubles differently than CVI. Therefore, functions
that return these types cannot be called from CVI if they are defined in Borland, or vice
versa. The exceptions are the ANSI C library functions that return doubles, which you
can call from within Borland compiled modules.
Note
This rule applies only to return values. You can use structs, floats and doubles as
output parameters without limitation.
00ProRef.book : 06chap04.fm Page 2 Monday, March 9, 1998 3:23 PM