Hacker News Reader

Comment by user3939382 | Hacker News Reader

I'm happy for them to branch out, but it is discouraging that once every 3-4 months I run into a bug or feature I need, google it, and am taken to a bug report in their tracker from 5-10 years ago that hasn't been touched.

View on HN

I don't know of any software projects of comparable size and complexity that doesn't have years-old bugs languishing in limbo. This is the normal state of affairs for a web browser, and it's the result of human-factor bottlenecks that aren't easy to solve even for well-capitalized projects. Large projects end up having subsystems that work reliably enough, such that their developers can afford to shift their attention elsewhere. Over time various details fall out of working memory. You get to a point where there is a high up-front cost to making any significant changes to the subsystem because it requires a significant investment of time for people to acquire (or re-acquire) the degree of familiarity that is needed to make such changes comfortably. If this cost can be amortized over a large backlog of feature requests and bugs, so it can keep someone busy for a while, then it is worthwhile. But it often isn't.

Upvoted for the 1st sentence.

A primary software at work is $1,500 to $10,000 a seat (depending on extensions) plus 10% annual maintenance. Our department annual fee is ~$70,000. I have a long list of "vote up this enhancement request or bug to get it fixed" that goes back almost two decades. (It's hard to identify the oldest because the vendor has switched tracking platforms twice.)

Case in point: network ping spikes after using Geolocation that will persist until FF is closed:

https://bugzilla.mozilla.org/show_bug.cgi?id=1711854

I never understood what geolocation was supposed to be within the context of of a desktop browser. If a website asks for this information, and I approve, what information is transmitted that couldn't be determined serverside via geoip?

GeoIP tells you at best what city someone is in, and even then, for smaller towns, will usually report the nearest large city, which could be quite a ways a way. Not to mention it's often even just flat-out wrong. VPNs and other things distort where you are. Sometimes a website might need (or at least think it needs) correct, fine-grained location information.

I do suspect that most websites that ask for location information could do fine with GeoIP, but that doesn't mean there isn't any use for more reliable, fine-grained location data.

In my experience, it will only reliably tell you what country you are in.

Okay, but I never told Firefox my location, so what’s it doing that geoip can’t?

like Google and apple - they use exact locations of WiFi hotspots most of the time...

That’s not something a webpage can access.

It's something a webpage can access on it's backend (the database of ssid/bssid and their locations) and something your computer uses(the access points its seeing) to locate you in the lack of a gps receiver and saying yes to the browser seeing your location.

Browsers have geolocation api, webpage uses it. Then browser utilizes underlying OS capabilities to determine the location. On the phone it could use GPS, regular PC may relay on SSID of wifi networks, etc.

BSSIDs of nearby WiFi networks that are compared to lists sent by mobile devices with real GPS

I don't think the BSSIDs are sent using the HTML5 location service.

Instead your computer operating system sends the BSSIDs to another service, get the location back and then your browser uses this location service to send a lat,long to the requesting website.

See https://support.apple.com/en-au/guide/mac-help/mh35873/mac and https://support.microsoft.com/en-us/windows/windows-location...

And if I don't even have a wifi card?

Then you represent a trivially small fraction of users. My desktop doesn't have wifi either, but most modern motherboards come with a wifi card built in. Not to mention the majority of visitors to mainstream sites are on mobile and a good chunk of the rest are on a laptop. Also Firefox has 3% market share so we're talking about a tiny fraction of a tiny fraction of a tiny fraction.

I run Qubes unless I'm doing something very performance-oriented like massively parallel workloads. So Firefox doesn't know about the builtin wifi card, which has no antenna connected, because I don't use it and it's just in the way anyway.

So yeah I have a wifi card, technically, but also technically don't from the point of view of firefox or any other program on my computer.

Most? More likely than not you will not have such a card in a mainboard.

Yes, most. Look at X470 or z590 mobos that are 2 generations old. The majority have wifi though I personally don't see the point.

Random board: https://www.gigabyte.com/Motherboard/X470-AORUS-ULTRA-GAMING...

Lol… geoip has never once gotten the state right for me. Best it does is get the country right.

I think this is a gnome bug rather than an ff one, but it's pretty shit on my desktop. It reports absurd (<10ft) accuracy while being off by miles.

Edit: s/100ft/10ft after I remembered better

On the desktop, it might be limited in the sources it has. GeoIP is one source, of course, but that's often wrong or too coarse-grained (or distorted by things like VPNs). The BSSIDs of nearby WiFi access points can help make that better, but that depends on you being close enough to APs in its dataset. Most desktops don't have actual GPS hardware.

Of course. But the desktop reports both estimated location and estimated accuracy. I'm complaining about the inaccuracy of the accuracy, not of the location.

Most MacBooks have a GPS antenna, so they will return pretty accurate location. And having a consistent API is important too, not having to care if your code is running on mobile or desktop is pretty nice.

What? Zero MacBooks have GPS inside.

Are you sure they don't just use WiFi SSIDs to determine a location? I can't find any evidence to support macbooks having GPS.

Although I am unfamiliar with what is sent with geolocation, I do use a VPN, and allowing geolocation on a website does pull my correct location even when connected somewhere far away.

The geolocation API uses location services on your device. This doesn't rely on your IP and instead looks at nearby WiFi access points and uses a 3rd party service to translate them to GPS locations.

for implicit it is reverse-geocoded from your IP address

for explicit it is using device GPS and/or cell tower

either you are not actually connected to the VPN or the website is storing your consent to access GPS coordinates. I'm not sure which is scarier

to test the former go to a few different websites that tell you your IP address

for the latter, fully clear your browser's cookies and cache and retry. does it ask to access fine location?

another reason could be that you are logged in to e.g. google who already knows where you are based on historic searches, nearby wifi and bluetooth devices, etc.

Thank you for the response, most likely it is browser cookies, as I've checked my IP address has changed and that there are no leaks. I will try running from a different browser in incognito mode to see if it still works.

WebRTC leaks expose your real identity and location even when you use a VPN. You should disable it in your browsers.

Agreed, I disable WebRTC, and run uBlock Origin and Privacy Badger.

I've had a cross-platform issue for ELEVEN YEARS with Firefox randomly not copying text into the clipboard, even when I 'cut' it-- i.e. the text disappears on cut but paste pastes whatever was in the clipboard before. I cannot be the only person with this problem. And yet the bug persists, and invariably someone pops in and tells me it's down to the extensions I have installed, despite not having any of the same extensions now that I did in 2012.

As a user, Firefox feels like hobby project, but at least it's not a Google product.

Reply on news.ycombinator.com