RE: [vfs] why is FileSystemConfigBuilder.setParam() protected?

2009-05-29 Thread Mario Ivankovits
 This is one of my least favorite parts of VFS. To configure  
 FileSystemOptions you have to use the appropriate  
 FileSystemConfigBuilder. I can't tell you why it was done this way as  
 it precedes my involvement, but I've considered reinventing this more  
 than once. It isn't pretty.

The reason I have done it that way was simply to have type-safe configuration 
parameters.
I hate it to have just setParam(String, String) or something like this.

It also was necessary to being able to mix parameters for multiple file-system 
implementations as
1) your url might create a layered filesystem
2) you never know exactly which filesystem implementation is going to be used

So, I still like this approach.

For the setParam(String, String) approach VFS provides the 
DelegatingFileSystemOptionsBuilder which then allows you to use 
setConfigString/setConfigClass which accepts just strings and uses reflection 
to coerce it to the required type. This was created to populate the 
configuration e.g. from an ant script or a properties file.

Ciao,
Mario

-
To unsubscribe, e-mail: user-unsubscr...@commons.apache.org
For additional commands, e-mail: user-h...@commons.apache.org



Re: [vfs] why is FileSystemConfigBuilder.setParam() protected?

2009-05-29 Thread Benoit Maisonny

 For the setParam(String, String) approach VFS provides the 
 DelegatingFileSystemOptionsBuilder which then allows you to use 
 setConfigString/setConfigClass which accepts just strings and uses reflection 
 to coerce it to the required type. This was created to populate the 
 configuration e.g. from an ant script or a properties file.
   
Great! This is exactly what I was looking for. May I suggest updating
the javadoc of FileSystemConfigBuilder and insert a link to
DelegatingFileSystemOptionsBuilder? As the delegate does not implement
the abstract class, it's more difficult to find it. I propose the
following text:

pWhen the VFS type is known at compile-time, developers should use the
VFS-specific *FileSystemConfigBuilder to benefit from type-safe
configuration. For generic configuration independant from the VFS type,
use {...@link
org.apache.commons.vfs.util.DelegatingFileSystemOptionsBuilder}./p

Would be worth putting the same text in the javadoc of FileSystemOptions.

Cheers,
Benoit


-
To unsubscribe, e-mail: user-unsubscr...@commons.apache.org
For additional commands, e-mail: user-h...@commons.apache.org



[commons math] solving a n equations system with n unknows using commons math

2009-05-29 Thread Julien Martin
Hello,
I would be very grateful if someone could tell me which class or package I
should look into in order to solve a linear system of n equations with n
unknows.
Please advise,
Thanks in advance,
J.


Re: [commons math] solving a n equations system with n unknows using commons math

2009-05-29 Thread Luc Maisonobe
Julien Martin a écrit :
 Hello,
 I would be very grateful if someone could tell me which class or package I
 should look into in order to solve a linear system of n equations with n
 unknows.
 Please advise,
 Thanks in advance,
 J.
 

Using the version in the subversion repository, you should do as is
depicted in the user guide here:
  http://commons.apache.org/math/userguide/linear.html

If you use the older 1.2 version, there was no separate Decomposition
algorithms, but the LU decomposition was directly provided by the
RealMatrixImpl class.

Luc

-
To unsubscribe, e-mail: user-unsubscr...@commons.apache.org
For additional commands, e-mail: user-h...@commons.apache.org



Re: [commons math] solving a n equations system with n unknows using commons math

2009-05-29 Thread Julien Martin
thanks a lot Luc!

2009/5/29 Luc Maisonobe luc.maison...@free.fr

 Julien Martin a écrit :
  Hello,
  I would be very grateful if someone could tell me which class or package
 I
  should look into in order to solve a linear system of n equations with n
  unknows.
  Please advise,
  Thanks in advance,
  J.
 

 Using the version in the subversion repository, you should do as is
 depicted in the user guide here:
  http://commons.apache.org/math/userguide/linear.html

 If you use the older 1.2 version, there was no separate Decomposition
 algorithms, but the LU decomposition was directly provided by the
 RealMatrixImpl class.

 Luc

 -
 To unsubscribe, e-mail: user-unsubscr...@commons.apache.org
 For additional commands, e-mail: user-h...@commons.apache.org