After upgrading to Redhat 8.0 I have a number of unit tests failing (you can see this by checking out jakarta-poi from Apache per instructions here http://jakarta.apache.org/site/cvsindex.html and running "cd jakarta-poi;export CLASSPATH=;./build.sh clean compile test" or run the "site" target to see the stack trace/all failed unit tests in a nifty html report under build/docs/tests/junit). It appears that they are all centered around java.lang.String behaving differently. Could this be a result of the default encoding changing between 7.3 and 8.0? If so...from what to what?
Here is an example of something that has changed: http://cvs.apache.org/viewcvs.cgi/jakarta-poi/src/testcases/org/apache/poi/util/TestStringUtil.java?rev=1.5&content-type=text/vnd.viewcvs-markup public void testPutCompressedUnicode() { byte[] output = new byte[ 100 ]; byte[] expected_output = { ( byte ) 'H', ( byte ) 'e', ( byte ) 'l', ( byte ) 'l', ( byte ) 'o', ( byte ) ' ', ( byte ) 'W', ( byte ) 'o', ( byte ) 'r', ( byte ) 'l', ( byte ) 'd', ( byte ) 0xAE }; String input = new String(expected_output); StringUtil.putCompressedUnicode(input, output, 0); for (int j = 0; j < expected_output.length; j++) { assertEquals("testing offset " + j, expected_output[ j ], output[ j ]); } StringUtil.putCompressedUnicode(input, output, 100 - expected_output.length); for (int j = 0; j < expected_output.length; j++) { assertEquals("testing offset " + j, expected_output[ j ], output[ 100 + j - expected_output.length ]); } try { StringUtil.putCompressedUnicode(input, output, 101 - expected_output.length); fail("Should have caught ArrayIndexOutOfBoundsException"); } catch (ArrayIndexOutOfBoundsException ignored) { // as expected } } The problem is that upon construction of the string, the last byte 0xAE is dropped. There are other string based unit tests that fail but this is the most clear cut. (The others I'm not yet sure why they are failing) So all of these unit tests are currently running fine (all things being equal) on Redhat 7.3! In fact the nightly build runs on my yet-to-be-upgraded webserver. Now this is all under JDK 1.4.x -- Java 1.3.1x doesn't even run! Java 1.3.1 just does this: [andy@localhost jakarta-poi]$ ./build.sh clean compile <snip> compile: Compiling project core with Java 1.3, debug on, optimize off, deprecation on Compiling 285 source files to /mnt/u03/home/andyext/jakarta/2/jakarta-poi/build/classes error: compiler message file broken: key=compiler.err.sun.io.MalformedInputException arguments=null, null, null, null, null, null, null /mnt/u03/home/andyext/jakarta/2/jakarta-poi/build/java/org/apache/poi/hssf/record/MulRKRecord.java:66: org.apache.poi.hssf.util.RKUtil is not public in org.apache.poi.hssf.util; cannot be accessed from outside package import org.apache.poi.hssf.util.RKUtil; ^ /mnt/u03/home/andyext/jakarta/2/jakarta-poi/build/java/org/apache/poi/hssf/record/RKRecord.java:59: org.apache.poi.hssf.util.RKUtil is not public in org.apache.poi.hssf.util; cannot be accessed from outside package import org.apache.poi.hssf.util.RKUtil; ^ /mnt/u03/home/andyext/jakarta/2/jakarta-poi/build/java/org/apache/poi/hssf/record/MulRKRecord.java:172: org.apache.poi.hssf.util.RKUtil is not public in org.apache.poi.hssf.util; cannot be accessed from outside package return RKUtil.decodeNumber((( RkRec ) field_3_rks.get(coffset)).rk); <snip/> Notice the peculiar message about the compiler message file being broken. Now the odd thing is the RKUtil is public and accessible outside the package (and again, it still works from jdk 1.3.1x under RH 7.3). So I guess aside from this being just obvious to someone as to why something like this would happen between RH 7.3->8.0.. my question is "has the default system encoding changed" -- "where would I find it" and what did it change from -> to? Your help is greatly appreciated. -Andy -- http://www.superlinksoftware.com - software solutions for business http://jakarta.apache.org/poi - Excel/Word/OLE 2 Compound Document in Java http://krysalis.sourceforge.net/centipede - the best build/project structure a guy/gal could have! - Make Ant simple on complex Projects! The avalanche has already started. It is too late for the pebbles to vote. -Ambassador Kosh -- redhat-list mailing list unsubscribe mailto:redhat-list-request@;redhat.com?subject=unsubscribe https://listman.redhat.com/mailman/listinfo/redhat-list
