This is the mail archive of the
java@gcc.gnu.org
mailing list for the Java project.
Re: gcj -bootclasspath doesn't work anymore?
Tom Tromey wrote:
Actually, I was thinking of [a javac-like script] that would also ignore
> javac arguments we don't support: -X, -source, -sourcepath, -target,
> etc.
My inclination would be to add support for those into gcj itself.
One question is: what should we call a javac wrapper? Should we
call it javac? I don't think we want to do that, at least by default.
Certainly gcc make install should not install $bindir/javac. If an
OS has an "alternatives" mechanism, then I think 'gcj -C' or a
wrapper script should be a valid alternative for javac, but we
should not preclude both being installed.
In my Makefiles for Kawa, I used $(JAVAC). JAVAC defaults to "javac",
except if you configure --with-gcj, in which case it defaults to
"gcj -C". I think this works well, and minimizes confusion.
So what is the advantage of a shell script that is argument-compatible
with javac, that invokes gcj -C, yet is installed under a different name
than javac? I don't see it. It adds an extra layer of indirection and
slowdown, plus its an extra script that needs to be maintained,
compared to just adding the desired functionality into jv-scan.c.
G++ used to be a shell-script, but we moved away from that.
The script Michael Koch sent does this.
I guess I missed that.
--
--Per Bothner
per@bothner.com http://www.bothner.com/per/