Date: Thu, 21 Jun 2001 18:04:27 -0700
From: ###@###.###
To: ###@###.###
Subject: Possible nio bug?

Look at

    /net/jano/export/disk20/GammaBase/Bugs/4386959/
        linux_i386-JDK-1.4.0-beta_refresh.20010621.1725.1

and compare it to the other

        linux_i386-JDK-1.4.0-beta_refresh.20010621.1725.*

files in that directory. (Not any that may still be in
progress when you get there.) That one says

        ...
         2,273 243 9 10,216K
java.lang.IllegalArgumentException
at java.nio.Buffer.position(Buffer.java:208)
at java.nio.charset.CharsetDecoder.handle(CharsetDecoder.java:362)
at java.nio.charset.CharsetDecoder.decode(CharsetDecoder.java:327)
at java.io.InputStreamReader$CharsetFiller.fill(InputStreamReader.java:368)
at java.io.InputStreamReader.read(InputStreamReader.java:511)
at java.io.Reader.read(Reader.java:103)
at Engine.copyFileFromReader(LoadTest_pbk.java:397) 2,283 244 9 8,826K

at Engine.loadHTMLInput(LoadTest_pbk.java:347)
at LoadTest.DoMerge(LoadTest.java:260)
at MergeThread2.run(LoadTest_pbk.java:492)
         2,291 245 9 7,694K
        ...

in the middle of what looks like a perfectly good run.
The runs on win32_i486 and solsparc don't seem to do that.
The other linux_i386 runs don't seem to do that. Is it
something we should worry about, or just put it down to
randomness?

Everything I know about the bug is in that directory or the
corresponding bug report. The source code is in there. It
used to throw OutOfMemoryError, but we seem to have fixed that.
It's no where near running out of memory at the point it gets
that exception. The difference between LoadTest.java and
LoadTest_pbk is that I took out an annoying "\r" from the
output at line 574.