On Apr 25, 3:53 pm, [EMAIL PROTECTED] (Michael Erskine)
wrote:
> On Tuesday 24 April 2007 15:35:04 Bob Hunter wrote:
>
> > But I am just repeating myself. So is PAR ... up
> > to par with technology? If not, is this from me a new
> > idea? If not, why it has not been developed already?
>
> Hi Bob, I produce Perl applications for both Unix and Windows platforms; I
> package using PAR on the target platform and it works very well.
>
> What you are suggesting is not a new idea and I hazard that it hasn't been
> developed already because those that have tried have failed and reverted to
> the existing method of packaging on the target platform that works perfectly,
> and it works right now.

You "hazard", that is you have no evidence of it. My reply is, that if
I run
gcc on linux using flags for windows, I get a windows executable. It
is
easy, so how could they possibly have failed? Pure perl is cross
platforn,
C can be compiled on any platform for any other platform, so it is
possible
for PAR to detect X flagged modules, get their source, compile them
for
the target platform, and pack the binary in the package. It is easy.

> On the subject of system calls: I often construct a system call at runtime
> according to some input that is unknown at time of packaging; hell, sometimes
> I decide on a library to use at runtime! And in such cases, how can PAR
> possibly work out what that system call is going to be?

I do not see the need to work out what system calls are made. Just
pack
cygwin.dll, and trigger the calls on it. That is what cygwin is made
for.
The solution is already there, and it is ready and easy to use.

> I think you'll find that it ain't worth complicating a simple tool (pp,
> simple? :) ) by adding incomplete and flawed features when a bit of pragmatic
> common sense and knowledge of the application you're packaging will suffice.

Picture the situation where you want to make a PAR that works on
windows,
but you do not have a windows machine (neither hard, nor virtual). PAR
can
generate true cross platform executables via gcc and cygwin. They are
there, and ready to use. All PAR can do is to use them, and it would
be
a great new feature.

Bob

Reply via email to