Karl Guggisberg <[email protected]> writes:

>>  maybe we should add a comment that uploads should be done in 1 transaction 
>> only until this bug is fixed.
> This wouldn't really help. Even if data is uploaded in one "transaction" 
> (i.e. with exactly one upload into one changeset before the changeset is 
> closed), we may run into the problem if the user cancels the upload.
>
> Anyway, a warning message couldn't harm, but I'd display it when the user 
> cancels an upload. It should read something along the lines of:
>
> * if the upload included new objects
>    "You canceled the upload of new objects. If you retry the upload the same 
> objects could be uploaded again. Before the next upload
>     please update your data and run validator for duplicate nodes."
>
> * if the upload didn't included new objects
>    "You canceled an upload. If you retry the upload you might create a lot of 
> conflicts because your local dataset might be out of sync
>     with the data on the server. Before the next upload  please update your 
> data and resolve any conflicts."
>
> But both these warnings aren't really helpful because in both cases you may 
> end up with strange data and/or a lot of work (resolving
> hundreds of conflicts, for instance). We will probably have to implement the 
> workaround described in #4401 and hope for an improved
> API/protocol in API 0.7.

If an upload fails, times out, gets canceled, or whatever JOSM could
examine the changeset and see what matches the data to be uploaded.

Matthias

_______________________________________________
josm-dev mailing list
[email protected]
http://lists.openstreetmap.org/listinfo/josm-dev

Reply via email to