Esp. your second paragraph I have forwarded to the network administrators. I think they have actually forgotten to give access on the sharing level.

What still intrigues me is how this API stores it. Afterall, all who share the file must be able to access that information.

>If I recall correctly, RLOCK() and FLOCK() do not actually modify the file -- so it doesn't matter if these files are effectively read-only. If I'm not mistaken those function use API calls to perform the lock -- so the locking is handled at the OS level. I vaguely recall that there was a byte-range locking scheme available with the DOS API (and I've got a vague recollection somehwere that negative values were somehow involved -- which implicitly results in a 2GB file size limitation).
>
>If I'm not mistaken, DOS-level file attributes and Windows security rights are completely separate -- so you can't really look at the DOS-level file attributes to determine if you've got write access or not. Files could have R/W attribute at the DOS level, but you might not have access rights to modify the file. Furthermore there is also the difference between access rights on the security level and those that are in the sharing level. The security rights may give you write access, but if the sharing not, then you don't have write access if the file is being accessed through a share (however if you're accessing it as a local file then it's OK). If you grant a user write access at the sharing level, but forget to give it in the security level, then the file is effectively read-only.
>
>>Read Only attribute on Folders has no meaning under Windows
>>
>>>I have now checked this and it is indeed the cause: The folder is ReadOnly.
>>>
>>>Interesting AND troublesome. I now realize that my RO/RW testing algorithms have a flaw: They do not test the folder's RO attribute. In my opinion it is a bug, or at least a quirck, of VFP. How on earth is it possible that rlock() returns true if isreadonly() is also true?!
>>>
>>>It makes me wonder where VFP stores the information that the user has locked a record.It can't be within the table, it can't be local, it can't be in a remote folder, it can't be in local memory. Who knows more about the how-and-where of storing lock-information?