[OSM-dev] Dev&svn accounts

[Shaun McDonald]

On 7 Mar 2008, at 14:01, Tom Hughes wrote:

> In message <9ACC0F05-9D32-4A0A-B097-8C0A8526BCF5 at shaunmcdonald.me.uk>
>        Shaun McDonald <shaun at shaunmcdonald.me.uk> wrote:
>
>> I already have a patch here (not created a ticket on trac or it yet)
>> to stop a user from uploading another file with the same name. Before
>> I can go on and implement: upload an archive and only import the  
>> files
>> that the particular user hasn't uploaded already; I would need to  
>> bash
>> some implementation ideas from someone who has more knowledge of
>> rails.
>
> I may be being dense here, but what's the problem with a user
> uploading a second file with the same name?


All my traces currently have a unique name as produced by the gps  
[software], thus I see it to be unlikely to be an issue.
Just to clarify the restriction is on a user-by-user basis, not system  
wide. So every user can upload a trace with the same name. However I  
cannot upload a trace to my account twice.
In database terms it would be a unique key on two columns (user id,  
trace name), rather than just one column (trace name).

The problem I currently have is that I have uploaded some of my  
traces, but don't want to upload lots of duplicate points. Currently  
the only way to prevent duplicates is to manually go through your list  
of traces to see if you have uploaded it already. This is very time  
consuming. Why not get the computer to do that difficult and time  
consuming job?

>
>
> As for the second part of your scheme, that sounds like it will
> encourage people to constantly reupload their entire trace
> collection and leave us to fish through and find anything new
> and I don't see why that is a desirable thing to have.

I can see why you wouldn't like this, due to the extra duplicate  
processing.
All my traces are in monthly folders, so it would allow me to upload a  
month at a time. It could be extended so that a text description file  
is included to tell you about the traces. This would then be faster  
than uploading the traces individually.

Shaun