>Here is my problem. Please, sit down and read carefully (it is very messy!).
>
>- I use VFP 5.0a French.
>- I have a table with 700 records that represent differents products.
>- An input form has a textbox to get the product number and a combobox with the product description.
>- The textbox and the combobox both have the same RecordSource.
>- It worked great until a corrupted record appeared. Even then, only the corrupted record did not work. All other recors worked well.
>- When a corrupted record is used, a message like « ALIAS '' NOT FOUND » OR «
IS NOT AN OBJECT » or even a «PAGE FAULT» is given on the READ EVENT. The message may be different.
>
>- I tried to repair it by going on the record, SCATTER MEMVAR MEMO to get information in memory, changing the product number (primary key), adding a new record, GATHER MEMVAR MEMO to get back the information from memory.
>- When I do that, the new record works well until I delete the renamed record (The old one).
>
>- I looked at the table in Hexadecimal to see if there was differences between the good and the bad record: I did not found anything (let be franc: I am not an expert in hexadecimal!).
>
>- I tried to reconstruct my table by creating the structure from scratch and then to APPEND FROM the old table, to execute an SQL query and to direct the result in a new table, and many more, but the corrupted record always followed.
>
>- I then realized that if I set the DELETED OFF, the corrupted record works well. The problem is that the entire system is developped to function with DELETED ON.
>
>- I arrive to the following conclusion:
> 1) I have more corrupted record than I tought;
> 2) For a record to work, some other record MUST be there and some other MUST NOT be there;
> 3) It looks like I will not be able to repair this problem without recreating the entire table by hand. Even then, I have no garanties that the problem will not return.
>
>Can you help??? It would be greately appreciated! I am working to find a way to publish an exemple. It might help to resolve this weird problem.
Jacques,
Big problem, little data. It seems to me the real problem is not data corruption (in my life I only saw this type of resistant corruption once and even it was corrected by a new structure copy - only difference I exported to SDF and appended from there).
- Does a test copy of form work seamlessly with a different test table (free or in a different DBC).
- While textbox and combo both have same recordsource, what buffering is set, are transactions in effect, what is rowsourcetype for combo etc etc
- Data session private or not.
- Triggers ?
- Type of combo and is textbox RO or RW.
In textbox you enter an ID and perform a seek ?
I think I'm increasing the mess :(
Cetin
