libgcc_s.so.1 and libgcj.so.4

[Andrew Haley]

Erik Poupaert writes:
 > 
 > >>>>> Why not?
 > 
 > Because it would be an enormous coincidence that someone else would also
 > need the versions in 3.3/20030414. By the way, I may very well upgrade to
 > next snapshot, if it provides substantial advantages/bugfixes. I don't know
 > what RedHat 10 will distribute; probably that will depend on the progress in
 > 3.3 and the timing of the RedHat 10 release. And also, I must keep an eye on
 > other distributions and non-rpm installations too.
 > 
 > >>>>> If you're deployong on Linux, I assume you'll either be using some
 > >>>>> sort of package manager like RPM or you'll be shipping source.
 > 
 > Since I've licensed the applications under the GPL, I'll be shipping both.
 > As a convenience, I think that providing the binaries too, for both win32
 > and linux, will be appreciated. But then again, developer-type downloaders
 > would probably be more comfortable with recompiling from source. But
 > developers are not necessarily always the primary audience. I'd like to
 > reach non-developer types too. As a matter of fact, I have no idea how many
 > people will actually be interested ...
 > 
 > I'm planning indeed to use RPM, but not in a way that it depends on other
 > RPMs, which in their turn depend on other RPMs.

Well, it's your application and so of course it's your call.  For me,
the advantanges to managing shared libraries and dependencies with RPM
are overwhelming.  However, gcj is different from most libraries in
that it's not stable with regard to binary compatibility, so you do
have to have the right version.  Once we get libgcj binary
compatibility fixed, this problem will Go Away.  This doesn't apply to
libgcc, where there really isn't a reason to do anything other than
depend on a sufficiently up-to-date version.  However, it's very
likely that OS vendors haven't yet shipped libgcc 3.3...

 > That story is really not attractive. By the way, self-contained
 > RPMS go a long way in alleviating the "dependency hell" everyone is
 > complaining about -- and with good reasons.

 > >>>> Nonetheless, you can do this by invoking the linker by hand.
 > 
 > How would I need to invoke "ld", in order to obtain equivalent results to:
 > 
 > gcc --main=MyMainClass -o myexe myarc1.a myarc2.a myarc3.a -lmylib1 -lmylib2

gcc -v --main=MyMainClass -o myexe myarc1.a myarc2.a myarc3.a -lmylib1 -lmylib2

Andrew.