IIRC 28 was a flag field that flagged compound index, memo field addition and also database addition -
had old stuff barfing on vfp compatibility early on coming from libs offering dB3+ file compatibility in C and Modula2

Modula2 under OS/2 & extended DOS more years ago than byte offset - sigh...


>>>Just wanted to let you all know that I finally solved this. I was able to figure out that it only happened when storing into a memo field in the error log and only after the application’s reindex utility indexed the errorlog table. I instead recreated the table, and added the indexes manually, and all is well.
>>>
>>>I suspect it means my client and I need to spend some time understanding what was different about my manual indexing than what the code does, but in the meantime, I’m not able to get errors without crashing out of VFP.
>>>
>>
>>And a little more digging has turned up that the reindexing code for the application clears byte 28 of the DBF header, which for a VFP-created table (but not for an FPW-created table, which explains why I'm seeing it, but others aren't), eliminates the indicator that there's a memo field. We'll be cleaning that up.
>
>Wasn't it supposed to be just the lowest bit („has cdx“), which needs to be touched (set to zero before creating indexes from scratch)?