RE: [mdb-dev] mdbtools-java: thanks!, status?, bug reports

[Calvin S. <cal...@ho...>]

>First of thanks to whomever did the Java port, I believe it was Calvin 
>Smith.

Yep it was me.  Thanks for the report.  good to hear someone is using this 
stuff
Bug 1)
  looks like a bug to me, just fixed it and checked it in

Bug 2)
  that bug may or may not have been in the C code.  I have found other 
places where the C code was wrong bug did not segfault.  can you try the 0.5 
version of the C code and see if it's also broke there?  keep in mind the C 
code may return the correct result but still be corrupting memory

Current plans are to sync the libmdb package to current cvs thereby picking 
up new feature and bug fixes.
JDBC driver works well for common cases so it tends to get less attention 
until someone like yourself reports the bug and / or offers a patch

Thanks again

----Original Message Follows----

First of thanks to whomever did the Java port, I believe it was Calvin 
Smith.  I need to read data from an mdb file on a Linux system via a Java 
servlet, and this project (almost) allowed me to do that.  I first used the 
libmdb package ala mdb_export, but I found the test code that does 
DriverManager.getConnection("jdbc:mdbtools:" + filename), which 
worked well.  Even though very little of the java.sql interfaces are 
implemented, we have Statement.executeQuery and ResultSet.getString, which 
are sufficient for my purposes.  Thanks again!

Other posts indicate that the Java port is a bit out-of-date, so perhaps the 
memo bug I describe below has been fixed in the C version.  From the libmdb 
code, it looks like some manual intervention is required to turn the C into 
Java, so perhaps this is why.

I am not part of the mdbtools team and have not participated in any such 
project, so I will describe two bugs I found and apologize for not fixing 
them.

1) Java bug (crucial, but trivial): in DataResultSet.getString(String), the 
method loops through the column names to get the index to call 
DataResultSet.getString(int), which does the real work.  
DataResultSet.getString(int) correctly accounts for the fact that 
ResultSet.getXXX(int) method number the columns from 1, as does SQL.  
However, DataResultSet.getString(String) does not account for the offset.  
Line 168 "return getString(i);" in the for loop should be return 
"getString(i + 1);".  (I'm not sure what the plans are for the 
Java wrappers, but obviously a table from column names to indices in 
DataResultSet would be better.)

2) I'm getting various index out of bounds errors from 
mdbtools.libmdb.Data.mdb_memo_to_string that are preventing me from reading 
tables that have a column of type "memo".  I can't even query the 
other columns because the current implementation reads the entire tale into 
memory, no matter what columns are in the query (right?).  I'm getting 
negative values for the string length and a value > 4096 for row_start, 
which is used as an index into MdbHandle.page_buf.



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
_______________________________________________
mdbtools-dev mailing list
mdb...@li...
https://lists.sourceforge.net/lists/listinfo/mdbtools-dev