On Mon, 14 Jun 2010, Yegor Kozlov wrote:
(1) REL_3_7_BETA1 depends on ooxml-schemas-1.0.jar
If I do clean checkout and then "ant dist" then the compilation fails,
however if I run the compile-ooxml-xsds target first then the compilation
complets OK.
Looks like a missing dependency on one of the ant tasks, should be fixed
in r954507.
We should clearly distinguish: ooxml-schemas-1.0.jar is JDK 1.4 and
ooxml-schemas-1.1.jar is JDK 1.5.
The switch to compiling the ooxml schemas as Java 1.5 was made in late
2009 by yourself, or so svn blame tells me...
------------------------------------------------------------------------
r882927 | yegor | 2009-11-21 15:30:54 +0000 (Sat, 21 Nov 2009) | 1 line
included poi-ooxml-schemas and poi-examples in the release cycle, also
misc improvements in build.xml: merged scratchpad and main examples,
improved cross-module classpath dependencies, removed obsolete targets
------------------------------------------------------------------------
Quite happy to go with whatever you think about the naming though, since
it was your change! I've tweaked the docs slightly.
(2) build of REL_3_7_BETA1 is locale-sensitive. There are two unit tests
that fail if the sources are compiled in the ASCII encoding. I confirmed
the problem on Win7 and Solaris.
I'm surprised we didn't set the encoding before, but thanks for fixing it!
(3) Compilation against ooxml-schemas-1.1.jar completes with 124 new
deprecation warnings.
Actually, all calls of kind xmlbean.getXXXArray() are deprecated in favor of
xmlbean.getXXXList().
I've been seeing these warnings for months, but not thought much about
them. Anyone who has a svn checkout, who since late november has done an
"ant clean" will have been running on the Java 1.5 version of the schemas.
I think it would be good to clean up the warnings, but I don't see an
urgent need to do it before the beta, probably only before the final
I'm fine with using ooxml-schemas-1.1.jar and JDK 1.5 syntax but we
should adjust our codebase and test-drive it for at least one month
before cutting release.
See above, I think quite a few of us have been trying it for a while
without realising!
Another point is the recent research performed by Bruce Alcock
([email protected]) - improper calls of setXXXArray() can
significatnly affect performance. I made a number of improvements in
XSSFSheet and XSSFRow and would like to research performance of setXXXList()
vs setXXXArray()
There's no such thing as setXXXList - it still seems to be setXXXArray.
AFAICT, it's only the get methods which switch when you compile for Java
1.5, not the setters
I'm tempted to do a new set of beta 1 artifacts and restart the voting.
Would that work for you?
Nick
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
