Manualshelf

H06.09 Software Installation and Upgrade Guide

ManualsBrandsHP ManualsServerHP Integrity NonStop Release and Migration
61626364656667686970
H06.09 Installation, Migration, and Fallback
Considerations
H06.09 Software Installation and Upgrade Guide—544316-002
2-23
Compiler Default Change for Object Code
Generation
°
Users whose password contains embedded spaces must request a new
password.
Compiler Default Change for Object Code
Generation
Prior to the H06.08 RVU, the native C/C++, COBOL, and pTAL compilers generated
preemptable object code by default. Beginning in H06.08, the compilers generate non-
preemptable object code by default. Non-preemptable object code is more efficient
than preemptable object code and results in faster compilation and execution.
Preemptable object code is required only when building DLLs that provide preemptable
symbols. The H06.08 versions of the compilers provide a command line option to
specify to specify whether or not preemptable object code is generated.
Because of the performance implications, developers should specify the option to
generate preemptable object only when building DLLs that provide preemptable
symbols.
Default Floating-Point Type Change
Beginning in the H-series RVUs, the default floating-point format used by the C/C++
compiler is changed. In G-series, the default is Tandem floating-point format; on
H-Series, the default IEEE format. The two formats are incompatible; an application
must consist of either all Tandem format modules or all IEEE format modules. Certain
HP-supplied libraries use the default IEEE format. If user programs also use the default
format, there is no problem. However, any user program that explicitly specifies the
option to use the Tandem format will encounter a linker error when attempting to load
an HP module that uses IEEE format. The recommended workaround is to compile
applications using IEEE format.
Providers of DLLs should also be aware of this default change, and take steps to
provide libraries that are compatible with both floating-point formats. This can be
accomplished in the following ways:
•
Write the library to be floating point neutral
•
Provide two versions of the library, one in Tandem format and one in IEEE format.
Compiler Host Compiler Name Option Name
TNS/E Guardian CCOMP, CPPCOMP,
ECOBOL, EPTAL
GLOBALIZED
TNS/E OSS and
Windows
c89, ecobol -Wglobalized
Windows eptal -globalized