Valery was maybe a bit to strong in his statements. You both
are right.
ROOT is:
1) A set of OO frameworks that a programmer can use to build
applications. The frameworks are available as a set of header
files and libraries. Use and/or inherit from the ROOT classes
and link with the libraries to build your application.
2) An interactive standalone application, root (or root.exe),
which gives access to all ROOT classes via a command line
interpreter or GUI. The root application can be extended at
run time by loading shared libraries of user code.
For the application you describe 1) is what you want. However,
often it is more simple to consider 2). Especially when doing
data analysis where you will be changing your analysis code
frequently. In that case creating a shared library containing
your classes and dynamically linking it to root(.exe) and
then using the interpreter to call your code will be much
more efficient (see in $ROOTSYS/test the libEvent.so library
and the eventload.cxx macro). Maarten Bruinsma's case would
probably benefit from using 2).
Cheers, Fons Rademakers.
Ole Streicher wrote:
>
> Hallo Valery!
>
> Valery Fine writes:
> > One needs VERY STRONG REASON to make his OWN ROOT-like application.
>
> So, then I don't understand what "root" really is.
> Up to now I thought that it is a kind of library (like hbook/hplot or
> so) which can be used within my programs, and the "baisc program" is
> called "root".
>
> F.e., I think about writing a monitor program which gets histograms
> from a DAQ via network and then simply presents them as the user like
> (with an automatic updating of the most important histograms and
> numbers).
> This program should be usable for people which didn't see root up to
> know (and which will be confused simply by the word "canvas" as an
> example).
>
> I thought that I can simply take the "root" library, write some own
> functions to get the histograms via net, write my own (specific)
> menus, enable all "standard" root functions I want and have my monitor
> application.
>
> Is this the wrong way? Or too difficult? How such a problem should be
> solved at the moment?
>
> Tschuessi
>
> Ole
-- Org: CERN, European Laboratory for Particle Physics. Mail: 1211 Geneve 23, Switzerland Phone: +41 22 7679248 E-Mail: Fons.Rademakers@cern.ch Fax: +41 22 7677910