Hi!

As some of you probably already have noticed, we have released V3.30A of 
IAR Embedded Workbench for MSP430. This is a new major release and as 
such it contains a lot of new and enhanced features. The main ones are, 
in my opinion:

* The compiler now produces about 3 to 4 percent better code than the 
last compiler, however your mileage may vary.
Among a lot of things, the Cross Call optimization (a.k.a. procedural 
abstraction and code factoring) has been greatly
improved.

 * Improved floating-point support for 64 bit doubles. In fact, the 
entire 64 floating-point package has been rewritten
from scratch in assembler -- the new library is two, three, four *times* 
faster (depending on the operation) than the
old library, and any other floating-point library I've seen for the 
MSP430. (This finishes a journey we started with
V3.20, in that release the 32 bit floating-point library had been 
rewritten in a similar fashion.)

* Support for MISRA C -- a subset of C that is used in the automotive 
industry to ensure that applications are more secure. (This is included 
in the full version only.)

* Version control system support for systems that follow the SCC standard.

* C-Spy support the OSEK Run-time Interface (ORTI). Based on this 
interface, C-SPY now provides sophisticated
debug awareness for any ORTI-compliant OS.

* The linker can now link raw binary images.

    -- Anders Lindgren, IAR Systems
-- 
Disclaimer: Opinions expressed in this posting are strictly my own and
not necessarily those of my employer.

Hi,

what about the qualifier restrict, warning for undefined operations,
Boolean type and values (<stdbool.h>) and complex numbers (and NaNs)
(<complex.h>)?
And is there a maximum portability mode so that only volatile variables
can be modified by ISRs?

Regards,

Rolf

msp430@msp4... schrieb am 12.05.05 13:08:55:
> 
> 
> Hi!
> 
> As some of you probably already have noticed, we have released V3.30A of 
> IAR Embedded Workbench for MSP430. This is a new major release and as 
> such it contains a lot of new and enhanced features. The main ones are, 
> in my opinion:
> 
> * The compiler now produces about 3 to 4 percent better code than the 
> last compiler, however your mileage may vary.
> Among a lot of things, the Cross Call optimization (a.k.a. procedural 
> abstraction and code factoring) has been greatly
> improved.
> 
>  * Improved floating-point support for 64 bit doubles. In fact, the 
> entire 64 floating-point package has been rewritten
> from scratch in assembler -- the new library is two, three, four *times* 
> faster (depending on the operation) than the
> old library, and any other floating-point library I've seen for the 
> MSP430. (This finishes a journey we started with
> V3.20, in that release the 32 bit floating-point library had been 
> rewritten in a similar fashion.)
> 
> * Support for MISRA C -- a subset of C that is used in the automotive 
> industry to ensure that applications are more secure. (This is included 
> in the full version only.)
> 
> * Version control system support for systems that follow the SCC standard.
> 
> * C-Spy support the OSEK Run-time Interface (ORTI). Based on this 
> interface, C-SPY now provides sophisticated
> debug awareness for any ORTI-compliant OS.
> 
> * The linker can now link raw binary images.
> 
>     -- Anders Lindgren, IAR Systems
> -- 
> Disclaimer: Opinions expressed in this posting are strictly my own and
> not necessarily those of my employer.
> 
> 
> 
> .
> 
>  
> Yahoo! Groups Links
> 
> 
> 
>  
> 
> 

Hi Rolf!

>  what about the qualifier restrict, warning for
undefined operations,
>  Boolean type and values (<stdbool.h>) and complex numbers (and NaNs)
>  (<complex.h>)? And is there a maximum portability mode so that only
>  volatile variables can be modified by ISRs?


We don't have support for the qualifier "restrict", yet. Until we
do, 
you have to go through the trouble of passing -Drestrict= to the compiler.

stdbool.h we have had for ages, the same goes for stdint.h. However, new 
for this release is inttypes.h. Complex.h, on the other hand, contains 
helper functions for the new primitive "complex" type, a type that we
do 
not currently support.

Floating-point NaN:s are supported by the low-level floating-point 
routines as well as all C-level routines (well, in DLib, that is, not in 
the legacy library CLib...).

Concerning the "maximum portability" mode, well we have a "strict
ansi" 
mode, but then you can't interrupt functions at all... I don't think 
that your suggestion seems that practical, but if you like please 
provide an example of when it could be useful.

    -- Anders Lindgren, IAR Systems

-- 
Disclaimer: Opinions expressed in this posting are strictly my own and
not necessarily those of my employer.