Version A.03.10 Features

Features introduced in the prior release, HP aC++ version A.03.10, are listed below. They apply to HP-UX 11.x operating systems.

The HP aC++ Online Programmer's Guide contains full documentation. (See Chapter 3 of these release notes for access instructions.).

The HP aC++ Online Programmer's Guide has been updated from HP CDE format to HTML format viewable with your HTML browser. For details and access instructions, see Chapter 3 of these release notes under Online Documentation.
Standards based features include the following:.
- covariant return types (except for covariant return types with multiply inheriting types)
- declarations in for and conditional statements
  Note: In the case of for statements, you must specify the
  -Wc, -ansi_for_scope,on option.
- Koenig lookup
  Note: You must specify the -Wc, -koenig_lookup,on option.
The -I header file option invokes view-pathing. This option overrides the default -I<directory> header file search path.
The +inline_level<num> option now defaults to 0, no inlining is done (the same effect as the +d option).
Additional options for verbose compile and link information:
- +dryrun - Requests compiler subprocess information without running the subprocesses.
- +time - Requests subprocess execution times.
- -V - Requests the current compiler and linker version numbers.
Huge data - Support for uninitialized, non-automatic data objects to a maximum size of 2^61 bytes for arrays and C style structs and unions.
Advanced options to support optimization of parallel code on HP9000 K-class and V-class servers:
- +O[no]autopar - Parallelize loops that are safe to parallelize.
- +O[no]dynsel - Enable workload-based dynamic selection of parallelizable loops.
- +O[no]loop_block - Enable [disable] blocking of eligible loops for improved cache performance.
- +O[no]loop_unroll_jam - Enable [disable] loop unrolling and jamming.
- +O[no]parallel - Transform [do not transform] eligible loops for parallel execution on a multiprocessor system.
- .+O[no]report[=<report_type>] - Produce a Loop Report
- +O[no]sharedgra - Enable [disable] global register allocation.
- +tm<target> - Compile code for optimization with a specific machine architecture.
Note: The Parallel Programming Guide for HP-UX Systems provides in depth information about code parallelization. Refer to Chapter 3 of these release notes.
Option +DO<osname> allows you to set the target operating system for the compiler.
The +Oinlinebudget=<n> option now works correctly.

Version A.03.10 Features

Technical documentation

» Table of Contents