MozillaZine

GPSWeb Extension for Mozilla Firebird Opens Possibilities for Location-Specific Web Services

Wednesday December 24th, 2003

Davide wrote in to tell us about GPSWeb 0.1.2, an add-on for Mozilla Firebird that adds a User-Location header to every HTTP request containing the user's current position as determined by the Global Positioning System. Sites can then use this data to provide location-specific services, such as supplying directions to the nearest restaurant. The extension requires the user to have some GPS hardware connected to their computer (Davide tested GPSWeb with the Garmin E-Trex) and the GPSd Java daemon, which parses the data supplied by the GPS unit, to be running.


#1 What?

by bugs4hj <bugs4hj@netscape.net>

Wednesday December 24th, 2003 7:55 PM

Reply to this message

There are databases with all IP ranges and their GEO location, so I'm not sure I understand the purpose of this project, please fill me in.

Don't you feel stupid for asking the nearest restaurant in your own hometown? Also, the places I use a handheld GPS don't have restaurants but again, maybe I missed something ;)

#2 Re: What?

by nonpareility <jbird3000@hotmail.com>

Wednesday December 24th, 2003 8:44 PM

Reply to this message

Think laptop/PDA/whatever with a wireless connection.

#3 Re: What?

by glazou <daniel@glazman.org>

Thursday December 25th, 2003 7:02 AM

Reply to this message

The nearest restaurant in your own hometown ? Geez, my home town, Paris, is quite wide and has thousands of restaurants. And the idea that the latitude/longitude is given by the restaurant itself through the Web and not by some other third party is really interesting. The current fashion in cellphones in Europe is digital photo. Next one will probably be geo-location using the european positioning system.

#5 Re: Re: What?

by bugs4hj <bugs4hj@netscape.net>

Thursday December 25th, 2003 10:42 AM

Reply to this message

The keyword here is "nearest" and that one can be located without any electronic device, even in Paris. Also, "the idea that the latitude/longitude is given by the restaurant itself through the Web" might be cool, but they don't have a website (yet). "Next one will probably be geo-location using the european positioning system." Cool, but say goodby to your privacy.

#10 Privacy?

by laszlo

Thursday December 25th, 2003 4:45 PM

Reply to this message

>>"Next one will probably be geo-location using the european positioning system." Cool, but say goodby to your privacy.<< What do you mean? As soon as you switch on your mobile phone, or better, as soon as you can receive a call, the company providing the call knows your position. Privacy and mobile phones don't go together well, this is the status quo already.

#11 Re: Privacy?

by bugs4hj <bugs4hj@netscape.net>

Thursday December 25th, 2003 6:20 PM

Reply to this message

"as soon as you can receive a call, the company providing the call knows your position."

True, but that info isn't available for the general public...

#12 Re: Privacy?

by bugs4hj <bugs4hj@netscape.net>

Thursday December 25th, 2003 6:33 PM

Reply to this message

"as soon as you can receive a call, the company providing the call knows your position."

True, but that info isn't available for the general public...

#15 Re: Re: Re: What?

by davide71

Friday December 26th, 2003 3:32 AM

Reply to this message

In fact GPSWeb gives the user the power to turn on/off the user location header only when and where he wants. I think this is a fair use of position information because the user has a complete control on them. On the other hand, the mapping between IP and location is a "big brother" stuff and at all events is the inference of a data (the position) from another data (the IP number) that today follows the rule "ISP assign x.y.z.u to location K" but in the future new IP protocols could arise where ISP cannot do that way.

#17 Re: Re: Re: Re: What?

by bugs4hj <bugs4hj@netscape.net>

Friday December 26th, 2003 3:54 AM

Reply to this message

Can GPSWeb be used without a GPS device? GPS hardware is still rather expensive, and especially when you're going to use it for this kind of applications only. There are more ways to obtain your GEO location so that would be nice.

There are plenty of other ways to find whatever you are searching for, if you're connected to the internet anyway, so I am still not convinced this is ever going to work. Its more likely that people have their cellphone at hand, and not a notebook and GPS device, just for getting directions to a restaurant.

Also, this will only work for mozilla based browers because that's a total no no.

#18 Re: Re: Re: Re: Re: What?

by davide71

Friday December 26th, 2003 7:20 AM

Reply to this message

don't know why my messages are posted twice ?!

>Also, this will only work for mozilla based browers because that's a total no no

There are other open source browsers that can be enhanced this way (konqueror for desktop and for mobile as well). Among the others, IE could be enhanced too, but it is not our target platforms. It is a diffuse opinion that Web-like browsers, cell phones and GPS are converging to a single appliance, thus it may work :)

#16 Re: Re: Re: What?

by davide71

Friday December 26th, 2003 3:43 AM

Reply to this message

In fact GPSWeb gives the user the power to turn on/off the user location header only when and where he wants. I think this is a fair use of position information because the user has a complete control on them. On the other hand, the mapping between IP and location is a "big brother" stuff and at all events is the inference of a data (the position) from another data (the IP number) that today follows the rule "ISP assign x.y.z.u to location K" but in the future new IP protocols could arise where ISP cannot do that way.

#4 Re: What?

by mattdm <mattdm@mattdm.org>

Thursday December 25th, 2003 9:01 AM

Reply to this message

IP->GEO isn't very precise. Not only are the databases incomplete, it's fundamentally difficult for them to get much better. Sure, you can guess that I'm in Massachusetts, maybe even figure out that I'm in Boston (even though my ISP's main POP here is located across the river in Somerville), but can you tell I'm in Brighton just west of Allston? (Not unless I've gone out of my way to tell you so.)

#6 Re: Re: What?

by bugs4hj <bugs4hj@netscape.net>

Thursday December 25th, 2003 10:45 AM

Reply to this message

"IP->GEO isn't very precise."

That's only true for the current public database, and that's not the one I am using . Oh, and I don't have to guess, trust me. Some SPAMMERS already learned this the hard way, I've put their personal info, including ID# on the web. Next please...

#9 Re: Re: Re: What?

by jgraham

Thursday December 25th, 2003 4:09 PM

Reply to this message

OK, so if I give you my IP address, you can tell me where I live? That sounds really unlikely to me. How does your magic database deal with DHCP? I guess my ISP might have a resonable idea of where I was (should they bother to look), but I'm pretty sure that information isn't publically avaliable. A GPS based system would solve this problem, although there would be privcy concerns with the location being broadcast withy every request.

#13 Re: Re: Re: Re: What?

by bugs4hj <bugs4hj@netscape.net>

Thursday December 25th, 2003 6:48 PM

Reply to this message

"I'm pretty sure that information isn't publically avaliable."

Yeah, that's correct, this kind of information isn't available for just anyone, and it better be save to protect your privacy. However, there are pretty good, but expensive, databases available for agencies that like to have a bit more control.

Btw, Merry Christmas, Jim.

#14 Re: Re: Re: Re: Re: What?

by mlefevre

Thursday December 25th, 2003 7:14 PM

Reply to this message

There's still no way IP can be as accurate. ISPs that allocate dialup IPs dynamically don't allocate or route IPs for small geographic areas. I'm more familiar with UK ISPs than US ISPs, but for most of the larger ones, the best you can do is an area of 60 to 100 miles. If the ISP doesn't have telecoms arrangements to get calls connected to the internet somewhere near the caller, then the IP address will tell you pretty much nothing useful - it only indicates where the internet connection is, which isn't necessarily close to the caller.

And that's without even bringing proxies and suchlike into the equation - there's no guarantee you can even get the IP of the user. Best case, it's still an order of magnitude worse than GPS. Worse case, you might not even guess the right country or even continent.

#20 hmmm

by mattdm <mattdm@mattdm.org>

Saturday December 27th, 2003 8:55 AM

Reply to this message

I think it's *more* likely that you're paying large amounts of money for information that isn't as good as you think it is.

#21 keep trying

by Down8 <down8@yahoo.com>

Saturday December 27th, 2003 10:55 AM

Reply to this message

That still isn't accurate all the time.

The easiest example is AOL - they are based in VA, so on occasion, a customer in say, ND, could be given an IP that would look like it's come from VA. When I was using dialup a few years ago, my IP was mapped to the closest major part of SBC, which showed me as being in San Jose, when in fact I was 3hrs from there.

So I reitterate the above point: "IP->GEO isn't very precise."

-bZj

#7 Re: What?

by davide71

Thursday December 25th, 2003 3:35 PM

Reply to this message

My counter question is: do you think user position (I mean accurate position not "I'm in NY") a totally useless variable in the computation of Web services? Possible answers are: 1) indeed useless! 2) 0,001% of Web applications could exploit user position somehow 3) useful. If your answer is 1 then our discussion is over, but in all other cases consider that a GPS costs few USD and the cost of developing a system that adds a User-Location header to HTTP requests is nothing. Most of the work is already done by Mozilla, we must listen for outgoing requests, add a header and that's all. The GPSd daemon is a NMEA string parser which has been developed outside the scope of GPSWeb. So why shouldn't we try to integrate all the pieces. I think applications will come, more clever than seeking the restaurant around the corner. Read you soon.

#8 Re: What?

by davide71

Thursday December 25th, 2003 3:40 PM

Reply to this message

My counter question is: do you think user position (I mean accurate position not "I'm in NY") a totally useless variable in the computation of Web services? Possible answers are: 1) indeed useless! 2) 0,001% of Web applications could exploit user position somehow 3) useful. If your answer is 1 then our discussion is over, but in all other cases consider that a GPS costs few USD and the cost of developing a system that adds a User-Location header to HTTP requests is nothing. Most of the work is already done by Mozilla, we must listen for outgoing requests, add a header and that's all. The GPSd daemon is a NMEA string parser which has been developed outside the scope of GPSWeb. So why shouldn't we try to integrate all the pieces. I think applications will come, more clever than seeking the restaurant around the corner. Read you soon.

#19 eh privacy??

by jilles

Saturday December 27th, 2003 4:08 AM

Reply to this message

I don't think I want to broadcast my location with each request. There's just no way I am going to do that. I want to disclose my location to authorized (by me) servers only.

You might want to look at the J2ME location JSR to get an idea of how to properly design a location service.

#22 My conclusion

by bugs4hj <bugs4hj@netscape.net>

Saturday December 27th, 2003 9:49 PM

Reply to this message

Sure, GPS position systems must be more accurate, or all military service equipment would be useless. Now, it is great to see yet another new add-on or extension whatever you want to name it but I still see a dark sky because of this:

a) privacy: most people, including me, like to keep their privacy as is - and this might not be the biggest obstacle but it sure it a big one.

b) type of browser: it only works with a mozilla based browser - the percentage of mozilla browser users is still by far outnumbered by MS IE.

c) hardware: you need a GPS device, or it won't work at all. - the percentage of people that a) use mozilla and b) have such GPS device and c) don't care about their privacy is what? 0.00001%

d) costs: it only works if web authors a) know about this and b) like to support anything like this and most important, c) their boss pays them to do so. - I don't think this is ever going to work, because there are plenty of websites, with tons of JS errors of no support for mozilla.

You still like to have directions, to a nearby restaurant? a) use your phone b) use a yellow guide c) be human and ask someone d) use Google, or any other good search engine, and hope for the best.

Oh, and web designers can use something called 'zip codes', because that seems to work very well, without the need to change anything, or don't they deliver your mail at home?

#23 Re: My conclusion

by davide71

Sunday December 28th, 2003 2:58 AM

Reply to this message

>a) privacy: most people, including me, like to keep their privacy >as is - and this might not be the biggest obstacle but it sure it >a big one.

privacy has not be addressed till now. But it could be done, for instance managing a list of trusted sites in the browser and sending user location only to them. Other solutions are possible and can be investigated. Remember that our project is really in its inception.

>b) type of browser: it only works with a mozilla based browser - >the percentage of mozilla browser users is still by far >outnumbered by MS IE.

It only works with mozilla NOW. I don't see the problem to implement the same in other browser. Any browser sends HTTP headers, am I wrong?

>c) hardware: you need a GPS device, or it won't work at all. - >the percentage of people that a) use mozilla and b) have such GPS >device and c) don't care about their privacy is what? 0.00001

Cost of GPS is dropping. You'll have GPS under your skin in few years (I'm kidding ...or I'm not?!). Let's see... a) use mozilla and other browser with HTTP headers. b) have such GPS c) willing to have location based services and able to handle their privacy with few clicks is what? 0.00001% mmm... if we consider 1 billion users it is at least 100 users. It is not a negligeable market:). But I don't think your estimate is very accurate.

>d) costs: it only works if web authors a) know about this and b) >like to support anything like this and most important, c) their >boss pays them to do so. - I don't think this is ever going to >work, because there are plenty of websites, with tons of JS >errors of no support for mozilla.

a) they'll know if it works. b) they'll like if it works c) already answered to that.

Best from Davide :)