[aspectc-user] New AspectC++ release

[Olaf Spinczyk]

Hi Sergio,

Sergio Queiroz wrote:
> 
> ---------- Original Message -----------
> From: Mariano Ceccato <ceccato at itc.it>
> To: Olaf Spinczyk <Olaf.Spinczyk at informatik.uni-erlangen.de>
> Cc: aspectc-user at aspectc.org
> Sent: Mon, 22 Aug 2005 13:56:51 +0200
> Subject: Re: [aspectc-user] Defining an Aspect Instantiation Model
> 
>>For this reason, I'm very interested on this new release, in 
>>particular I'd like to know the time lines for this new ac++ release,
>> and the expected maturity. Moreover, could you please comment a 
>>little bit on the expected evolution of the ac++ project in terms of 
>>its usability in an actual industrial setting?
> 
> I don't know if somebody answered this question just to Mariano, but I also
> would like to know when should be released the new version of AspectC++ and
> what should be the new features.
> 
> Sérgio

sorry for answering so late on this important question! We have been in 
discussions about the next release and have been unable to answer earlier.

The next release will be called 1.0pre1 and it will come this month. 
Here is the list of new features/fixes that have already been implemented:

-Fixes:

* fixed the match expression parser, which crashed sometimes.

* fixed several parsing problems reported by Mariano

* fixed several parsing problems, which we found in the ACE library

* fixed the HUGE problem of ac++ 0.9.3 that cause parse errors, when
   an aspect defined code advice for a class into which it also
   introduces new functions/members

-Improvements/new features:

* tjp->proceed() can now be completely inlined. Thus, the resource
   consumption of around advice is much better

* support for ... in template argument lists in match expressions

* support for the gcc language extension "statement expression", which
   is used in several header files on linux systems

Things we are still working on:

* improved template support in the parser (later instantiation)

* doxygen documentation of the most important Puma classes

* and finally a very IMPORTANT new feature: modified code generation
   patterns that solve the include cycle problem almost completely.
   "Almost" mean that in some case, i.e. introductions, which introduce
   cyclic dependencies of modules, you'll have to use a workaround.
   However, the good new is: there will be a workaround. A more
   convenient solution will be developed step by step.

Regarding the usability in an industrial setting, I don't want to 
promise too much, but ac++ 1.0pre1 will be tested with real-world code: 
for example with the ACE library (Adaptive Communication Environment), 
the open source web server MyServer, and the kernel of the eCos 
operating system.

If you are interested in tests with other code, don't hesitate to ask or 
commit yourself to BECOME A TESTER. My medium term goal is to have a 
table on the AspectC++ web site that shows on which code ac++ has been 
used and if it was successful.

For information about our other goals for version 1.0 refer to the 
roadmap page at aspectc.org (which I will soon update).

Best regards,

Olaf