Jess Holle wrote:
Max Berger wrote:
Jess Holle schrieb:
As of October of this year anything prior to Java 5 will be officially
unsupported by Sun except where you have a paid support contract with
them (including that for Solaris). Add to that the fact that Java 1.4
is so limiting for so many things and Java 5 is supported on even the
most odd ball platforms (except Java ME, but I don't see FOP on ME...)
I have to disagreee. AAMOF, java 1.5 is ONLY supported on a select
number of plattforms, mainly the ones Sun provides the JDK for (or is
licensed). Those are:
Solaris / Sparc
Linux / Intel
Windows / Intel.
OS X / PowerPC + Intel (licensed)
*BSD / Intel (through emulation layers)
There are some be specific IBM jdks (on IBM plattforms), but I have no
experience with these.
That's it. Anyone using any other plattforms (such as Linux / Sparc,
etc.) has to use either the blackdown JDK ( stuck at 1.4 ), or different
JVMs, such as Jikes, Kaffee, etc., ALL OF THEM stuck at 1.4 as well or
not fully compatible (such as gcj). I have Linux / Sparc as a work
machine, and have been looking for a way to run 1.5 code for quite a
while... OpenJDK being the most promising solution (although still not
working).
I guess I shouldn't have said "even the most odd ball platforms" but
rather all but the most odd ball platforms.
Java 5 runs on HPUX and AIX machines as well -- as well as on Solaris
x86. When you've got Windows, Mac OS X, x86 Linux, Solaris, HPUX, and
AIX covered, I'd say that's pretty good. The other platforms are, er,
odd balls.
The question is whether you hold everything else back for the odd balls
that seem unlikely to progress beyond 1.4.
A view from the outside.
With both HPUX and AIX, I imagine that there is another variable - which
version of the OS supports 1.5. AIX versions don't come for the price of
a Vista licence, and the same probably holds for HPUX. In addition,
there may be minimum hardware requirements for upgrades.
Talk to Chris about this. He has always been prominent in defending the
FOP interests of his employers, and seems to know a bit about these issues.
Wile we're on the topic...
Concerning the issue of retaining 1.4 compatibility by hamstringing the
use of 1.5 constructs: a truly, deeply, madly ridiculous idea. Go to
1.5, or don't. Simple. Don't complicate things with this notion of using
generics, but nothing else.
If (real) 1.5 is feasible, what about 1.6. There was a huge jump in the
language itself between 1.4 and 1.5, which is why it has taken so long
to wean the world off 1.4. There is not such leap between 1.5 and 1.6.
The survey should canvas 1.5 and 1.6. I'd be curious to see how many
systems support the first and not the second.
The decision for FOP should be based, not on the end-of-life for a
particular version, but on the pragmatic reality of 1.5 vs 1.6 support.
That's the second glass of wine for the evening.
--
Peter B. West <http://cv.pbw.id.au/>
Folio <http://defoe.sourceforge.net/folio/>
