[OSM-talk] syj: preview of a web site to store routes, need your opinions :)

arno arno at renevier.net
Tue Jul 27 21:48:10 BST 2010


Le mardi 27 juillet 2010, à 18:11:09 +0100, David a écrit : 

> Given you don't like the idea of incrementally saving the route, how
> about replacing Save with a button which says "Give me the URL" (or
> whatever) which has the effect of Saving and then popping up an
> overlay which provides you with the URL (and maybe offers to email
> it to you). Doesn't require an extra button then to get the info
> needed.

nice idea :)

> >>I think the need to login (or rather, create an account) is
> >>EXTREMELY offputting. It's the biggest single reason putting people
> >>off a website in general. I don't see why you need it - can't you
> >>store IDs and/or names without? (Or even encode the route in a URL).
> >>Like bit.ly for example.
> ...
> >
> >The main reason a login is needed is to prevents spam or database attack. What
> >happens when some random scripts send 10 http posts a second ?
> 
> You just don't allow it. Check the size is reasonable for the time
> taken (you can record when they started in the session). Maybe
> enforce a maximum overall size.
> 
> You could also check that the same IP isn't sending reasonable size
> requests at too frequent intervals and block that IP (record the IP
> with the database entry) if it does.

may be this could work. Or something like: one ip cannot httppost twice in 
less than xx seconds. The main drawback I see is that sometimes, multiple 
users share the same ip. For exemple, people in university, or mobile 
networks, or whatever. I don't known how to overcome this problem.

> Apart from that, why do you care if it is a robot, so long as it
> isn't overloading you?

The other point I see is that if some day I implement a search function 
(looking in routes title), the database can be cluttered with crap, and a 
search possibly won't yield a valid resut. But I think that's a minor point.

> >>Do you need a "Save" at all? Why not save it live as they do it, so
> >>all you have to do is harvest the URL at the point you are ready.
> >
> >I'm a big fan of live saving on desktop applications, but don't like it for
> >web pages: it induces a lot of http requests
> 
> Well one for every click or drag, but that's hardly a lot unless
> it's a robot doing it.

It's not a lot if you're on a decent network. If you're on a slow network, 
or a network that often fails, it's different.

> I don't think you need to know this. All you need to do is append a
> point with an Ajax request each time they add one. If they go away,
> so be it. Why not save with each click or drop? It's not a
> complicated request or a lot of data.

If you have a long route: either you send the whole path each time, and this 
can make big requests, even for a decent network, either you send only the 
added/modified point, and you have to take care everything comes in the right 
order, and that' a mess.

> The danger of not doing it this way is that people will press the
> back button expecting it to undo the last point added, and end up
> discarding everything they've done.

The danger of doing it this way is to have a path that is not saved, but user 
thinks it is saved. 

The other problem is that people "expect" a save button. I have tried to 
create web interfaces without a save/validate button, and a lot of people did 
not understood how to use it.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20100727/8f208ee8/attachment.pgp>


More information about the talk mailing list