Re: NetBeans free Forte works (was Re: Forte for java as java IDE--Yes!

[Marcus G. Daniels]

>>>>> "PJ" == pauljohn  <address@hidden> writes:

PJ So If my swarm configure statement did not have
PJ> --enable-subdirs, then the swarm stuff would get installed into
PJ> /usr/lib, not /usr/lib/swarm, and the LD_LIBRARY_PATH magic would
PJ> not be necessary. 

..but this is only in the case of RPM builds for Linux, because the
resulting directory will be a system library (/usr/lib) and it is a
place that the dynamic linker will be looking anyway.  The absence of
--enable-subdirs isn't something that will fix it in general (say, on
a Sun).  I think it is dubious to switch to /usr/lib for the sake of
Java, since some of the Swarm libraries have generic-sounding names
(e.g. libmisc.so) and there is a danger of name clashes. 

PJ> Mostly, I was curious why the java program requires the
PJ> LD_LIBRARY_PATH setting when objective-c programs do not. After
PJ> you answered, it hit me that the executable script in the build
PJ> dir of an objective-c program has libtool magic that does all that
PJ> stuff, and LD_LIBRARY_PATH is not needed because of that.

Right.  Obscure tangential information: in newer versions of libtool, the
script goes away and a feature of Linux executables (ELF) encodes the
pathnames for for dependent shared libraries into the executable
itself, _even_ when it is in the build directory.  So maybe when
libtool 1.5 is released there will be no more `wrapper' script.  (Now, the
development version of libtool has other strange problems w.r.t. to
Swarm, and so the version distributed in the Swarm snapshots is 1.3b.)


                  ==================================
   Swarm-Support is for discussion of the technical details of the day
   to day usage of Swarm.  For list administration needs (esp.
   [un]subscribing), please send a message to <address@hidden>
   with "help" in the body of the message.