DO NOT REPLY [Bug 19739] - ContinueRecord not copied by RecordFactory

[bu...@apache.org]

DO NOT REPLY TO THIS EMAIL, BUT PLEASE POST YOUR BUG 
RELATED COMMENTS THROUGH THE WEB INTERFACE AVAILABLE AT
<http://nagoya.apache.org/bugzilla/show_bug.cgi?id=19739>.
ANY REPLY MADE TO THIS MESSAGE WILL NOT BE COLLECTED AND 
INSERTED IN THE BUG DATABASE.

http://nagoya.apache.org/bugzilla/show_bug.cgi?id=19739

ContinueRecord not copied by RecordFactory

brett.knights@tanner.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|Copied Spreadsheet from     |ContinueRecord not copied by
                   |Japan can't be opened.      |RecordFactory



------- Additional Comments From brett.knights@tanner.com  2003-05-09 15:40 -------
I stripped down the spreadsheet until I had a very simple one that still 
wouldn't copy. 

It turns out the problem was in the handling of ContinueRecords. In four cases 
in my spreadsheet ContinueRecords followed UnknownRecords. UnknownRecord can't 
process ContinueRecords so whatever information was there wasn't being copied 
to my new spreadsheet. 

I could probably have modified UnknownRecord to suck any following 
ContinueRecords in but and recreate them on serialization but the 
UnknownRecords weren't anywhere near the record size limit so I have no idea 
why the ContinueRecords were there in the first place.

Instead I made the following changes to allow my spreadsheet to copy:
Added a method to org.apache.poi.hssf.record.Record:

    /**
     * Does this record want to process a Continue record?
     *
     */

    public boolean isContinueRecordAware()
    {
        return false;
    }

I overrode this method to return true in SSTDeserializer and SSTRecord.

Replaced line 218 of RecordFactory with:
                               if(last_record.isContinueRecordAware()){
                                  last_record.processContinueRecord(data);
                               }else{
                                  last_record = record;
                                  records.add(record);
                               }

Then I made a bunch of changes to ContinueRecord to allow it to serialize like 
UnknownRecord. I can diff these if you are interested in applying this as a 
patch.

I don't know if this is the best approach to handling ContinueRecords though.

One thought is to make Record ContinueRecord-aware and have it just keep a list 
of following ContinueRecords. Classes like SSTDeserializer and SSTRecord would 
expand their ContinueRecords as they do now but would also keep their 
ContinueRecord copies and would only toss their ContinueRecords(to be 
reconstitued if possible) if any of their fields were modified. 

Unmodified classes would automatically include their ContinueRecords on 
serialization.