Roy,
I'm tossing a Hail Mary at what is probably an inappropriate moment, but if I don't try, I can never get surprised.
Major client has had a horrible time with BDE since the Fall 2017 Windows Update. What had been an occasional Blob is Modified issue and an even rarer Index out of Date problem has turned into a several times a day nuisance. Each time, it takes everybody getting out of the program and one of the office workers running Rebuilder (a modified version of PdxRbd). Then it's somewhere between five minutes and an hour before it's everybody out again.
I am working hard at turning my 25 year old program into an NexusDB program, but we have to hang on to the BDE version for another few weeks (I hope it's only a few weeks). I am asking if YOU are aware of anything I can do with settings or whatever to stop this rampant corruption madness. I theorize it's an issue between BDE and Windows locks. But whatever the cause, I need to cope with it.
Today's EurekaLog Error report ends with wwRichEd.pas's CopyRichEditFromBlob erroring out at line 2620 rel 8. The thread had gone through my calling for a LocateDialog.execute. Moves through some windows calls to Locate's FindFirst and then FindMatch functions before going to wwFindMatch and ApplyMatch. Next, into wwTable's DataEvent handler. The last stop is in twwRichEdit's, DataChange, LoadMemo and then two hits in CopyRichEditFromBlob before the error break.
Since this has worked for DECADES in a Windows 7 environment with nary a problem that is big enough to trouble you, pretty sure there's nothing wrong with your code. But with the advent of Windows 10 workstations and that update in 2017, all heck has broken out.
IS THERE ANYTHING YOU, or any lurker, can think of that I can change settings-wise to let me allow BDE and Win10 workstations and Windows Server 2013 I think (might be earlier) live together for JUST a few weeks?
Am at my wit's end repairing the damage the daily corruption does, which means I AM NOT WORKING ON THE REPLACEMENT SOFTWARE. Feel a lot like Sisyphus.
Regardless of answer, just want to know how much I appreciate your software. Hope you and yours enjoy a Merriest of holiday seasons and have a Happy New Year! GM
Sorry, I don't have any useful info to offer. You could try some of the steps someone else shared (but I don't know if these relate to Windows 10 issues).
*******************
I assume you're using paradox format. I believe it means one of the memos in your *.MB file is corrupt. And most likely corrupted due to the dreaded write caching...
Taking the following steps will minimize corruption problems (though eliminate would be better wouldn't it?):
1. Set Local Share to TRUE in bde config. 2. Setup a DBIUseIdle in AppIdle call. 3. Put DBISaveChanges in the AfterPost event of EVERY table.
Seems like overkill -- but only after using all 3 have I had some semblance of data integrity.
As for fixing your corruption, download the TUtility exe off of Borland's site. It will allow you to edit your memo field again. You will lose the data in that field, however, but only for the records that are corrupt.
Hope this helps,
Jeff H
********************