File seek in wlst fails for lines longer than 8091 characters

Question

For a CSV file generated in WLST i want to update the header, the first line of the output file, when new metrics have been detected. This works fine by using seek to go to the first line and overwriting the line. But it fails when the number of characters of the first line exceeds 8091 characters.
I made simplified script which does reproduce the issue i am facing here.
    #!/usr/bin/python
    import sys
    global maxheaderlength
    global initheader
    maxheaderlength=8092
    logFilename = "test.csv"
    # Create (overwrite existing) file
    logfileAppender = open(logFilename,"w",0)
    logfileAppender.write("." * maxheaderlength)
    logfileAppender.write("\n")
    logfileAppender.close()
    # Append some lines
    logfileAppender = open(logFilename,"a",0)
    logfileAppender.write("2nd line\n")
    logfileAppender.write("3rd line\n")
    logfileAppender.write("4th line\n")
    logfileAppender.write("5th line\n")
    logfileAppender.close()
    # Seek back to beginning of file and add data
    logfileAppender = open(logFilename,"r+",0)
    logfileAppender.seek(0) ;
    header = "New Header Line" + "." * maxheaderlength
    header = header[:maxheaderlength]
    logfileAppender.write(header)
    logfileAppender.close()
When maxheaderlength is 8091 or lower i do get the results as expected. The file test.csv starts with “New Header Line" followed by 8076 dots and followed by the lines 2nd line 3rd line 4th line 5th line
When maxheaderlength is 8092> the test.csv results as a file starting with 8092 dots followed by "New Header Line" and then followed by 8077 dots. The 2nd ... 5th line are now show, probably overwritten by the dots.
Any idea how to work around or fix this ?

Answer 1

For the benefit of others, i got a workaround from Jeremy Gosling on a crosspost on stackoverflow.
I too was able to reproduce this extremely odd behaviour and indeed it works correctly in Jython 2.5.3 so I think we can safely say this is a bug in 2.2.1 (which unfortunately you're stuck with for WLST).
My usual recourse in these circumstances is to fall back to using native Java methods. Changing the last block of code as follows seems to work as expected :-
# Seek back to beginning of file and add data
from java.io import RandomAccessFile
logfileAppender = RandomAccessFile(logFilename, "rw")
logfileAppender.seek(0) ;
header = "New Header Line" + "." * maxheaderlength
header = header[:maxheaderlength]
logfileAppender.writeBytes(header)
logfileAppender.close()