>>>

>>>CREATE CURSOR test (long_field_name c(1))
>>>
>>>ALTER table DBF("test") add shortf c(1)
>>>

>>
>>The second line won't be of any help here, as I found out earlier today.
>>
>>Perhaps you implied to say that ALTER TABLE does not deal with cursors. However, it does alter a cursor also. The next command will work:
>>
>>

CREATE CURSOR test (shortf c(1))
>>
>>ALTER table test alter column shortf c(2)

>
>Actually, the second line in my example will work fine IF the cursor has a disk presence. The problem is that VFP won't write the cursor to disk until it has to, otherwise it will keep it in RAM and there is no DBF file.

And then the funny and good thing is that [ALTER table test alter column shortf c(2)] will also do the job if the cursor has no disk presence yet.

Talking about dbf(alias())... I'm of opinion that a cursor should ALWAYS have the physical representation in the tmp-map. It is now indeed only created after some (how many???) records were added. This creates a condition of uncertainty for the developer. I think there are a lot of places in our code where we assume the existence. One day such a place may cause an error. Who's fault is that then? I think it's essentially the fault of the developers of vfp. So, I'm not pointing to a bug (because the behavior is actually documented), but this is surely an item for the wishlist!