ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Summerwill, Bob" <>
Subject RE: C++ tasks
Date Fri, 12 Oct 2001 14:21:23 GMT

Having a C++ compilation task allows C++ or mixed-language projects to move
from makefiles to XML build files.  This has got all the same benefits as
Ant already gives to Java projects.  I think it would be a fantastic
addition to Ant.


-----Original Message-----
From: Russell, Mark []
Sent: 12 October 2001 15:01
Subject: RE: C++ tasks

Just think of the case where your trying to integrate JNI C++ code with Java
code, and you'd like to build them all with the same build system.  That is
really the main reason (in my mind) for a c++ task for Ant


-----Original Message-----
From: Les Hughes []
Sent: Friday, October 12, 2001 04:06
To: ''
Subject: RE: C++ tasks


Feel free to shoot me down in flames but doesn't VC++ have a perfectly good
build system in nmake? And for multithreaded speeded up stuff what about
clearmake (with ClearCase).

(Oh dear, what have I started....)



> -----Original Message-----
> From: Curt Arnold []
> Sent: 11 October 2001 03:50
> To:
> Subject: Re: C++ tasks
> > I'm hoping to begin work on this again soon (in the next 
> few weeks) but my
> > intent includes a substantial rewrite of the initial task.  
> Mostly because
> I
> > want to speed the task up so I intend to include multithreading
> capabilities
> > into the task.  For now though this is still in design with coding
> hopefully
> > to start end of month.
> Speed is a very good thing.
> >
> > If needed I can repost both Adam and My tasks to the list 
> or else email
> them
> > direct to you.
> I'd appreciate your sending them to me.
> >
> > One word of warning is that the linker part of my task does 
> not function
> > properly on Sun Workshop, other then that everything should 
> be working
> My primary interest is driving the Microsoft Visual C++ 
> compiler on Win32.
> I'd appreciate your thoughts on compile speed minimization.  
> If compilation
> is processor-bound and you aren't on a multiprocessor system, 
> I wouldn't
> think that multiple threads would gain you any performance.  Though a
> thread-per-processor would probably be appropriate.
> Definitely, I would think that you would want to minimize the 
> number of
> processes created (that is not exec a new process for each 
> .c[pp] file) and
> minimize the number times CL.EXE is loaded.
> Would the same optimizations be appropriate for gcc or other 
> C++ compilers,
> other platforms?
> Any thoughts on dependency checking?

View raw message