bma/ journal/ tags/ google

This feed contains pages in the “google” category.

Google OS considered harmful

Google have announced that they are releasing an operating system. What I want to know is: why?

Well, obviously, because they can, because it’ll make them money. Really, though, what good will it do? If anything, I think it’ll just be harmful.

Firstly, what they are releasing is not an operating system. At the very least, the kernel is Linux, and I’m willing to bet a chunk of the userspace is GNU, or some other third-party (uClibc/busybox, perhaps). “Google Chrome OS” is merely a windowing system on top of that — no better than Apple did with Mac OS on top of BSD/Mach.

Secondly, they claim they’ll open-source it “later”. Pardon me, but I take that with as big a pinch of salt as “Google Chrome will work on Linux real soon now!”. Of course, they’ll have to release the source for any modifications to GPLed code as soon as they release anything, but I don’t particularly trust Google to do anything sooner than it suits them.

Thirdly, this is yet another platform built around webapps; third-party applications will be web-based. This, of course, makes the platform absolutely useless without a network connection; I truly hate that idea. (It’s not that uncommon to have no connection to the internet; mobile phone signal on the train from Plymouth to Exeter is spotty at best, for example, and without a 3G modem you’d be relying on wifi, which is even worse for coverage.)

Building the platform around webapps, furthermore, just ties people into those webapps; if a computer has no capabilities of its own beyond using web applications, the user has no option but to use web applications. Sure, they don’t have to use Google applications, but that’s not really choice (“a choice of masters is not freedom”). People should be able to choose to use webapps if that’s what works for them, but they need to have the option of keeping their data to themselves, on a system that they control. A system that relies on webapps doesn’t allow this.

If nothing else, this has already been done: WebConverger is a minimal system based on Debian and Firefox, for exactly the same use case. It’s still not a use-case that interests me, but at least it’s done the right way — taking existing components, putting them together, and improving them to do what you want, rather than an entire NIH stack (a new windowing system? what good will that do?).

The announcement was made less than 24 hours ago, and of course my opinions may change. I somehow doubt it, though.

Edit: removed inaccurate reference to the iPhone; thanks, Ross.

Thoughts on Google Chrome

Google are due to release a new browser, called Chrome, tomorrow. So, here are my initial (i.e., pre-release) thoughts on it, based solely on what they’ve said about it so far.

  • Initial release will Windows-only; Mac OS X and GNU/Linux versions coming Real Soon Now. Nice to see that even Google, with their thousands and thousands of GNU/Linux machines, treat anybody not using Windows as second-class citizens. Hardly unexpected by now, but still irritating.
  • I’m glad to see that they’ll be releasing it as free software; hopefully, they’ve the sense to use the GPL or something, rather than a new licence of their own design.
  • I’m not convinced by the tabs-as-separate-processes thing; is there really an advantage over tabs-as-separate threads, that outweighs the additional overhead of separate processes? I admit that I don’t know enough about memory management to seriously evaluate this one, though.
  • They make a big deal about webapps, but say nothing of the client side; will there be anything along the lines of Firefox’s extensions API? Gem suggests supporting Firefox extensions directly, which may be difficult without XUL support; an equivalent API is a must, though.

Beyond that: we’ll see?

Update: um, today, apparently. About to try it out under WINE.

GoogleTalk isn't Jabber

So, everybody’s been singing the praises of Google for using an open standard, XMPP (also known as Jabber), for its Google Talk service. Open standards are, undeniably a good thing. When they’re conformed to, that is.

I, as a Jabber user but not a Google Talk user, can talk to most Google Talk users just fine. Their JIDs (Jabber IDs) are something like username@googlemail.com or username@gmail.com; when you connect to them, your non-Google Jabber server looks for googlemail.com or gmail.com and gets redirected to the right place.

The fun comes when somebody is using Google Tools (or whatever it’s called) to provide Google services from their own domain; their JID will be username@example.com. However, example.com isn’t a Jabber server. My suspicion is that the Google Jabber servers know that example.com is a Google Tools domain and redirect to themselves. However, no other Jabber servers have any way of knowing what server to use instead of example.com—the whole point of a JID is that it tells you what server to use for a given user, and there’s no other way of specifying.

Well done, Google. You win again.

GMail cockup

Well, somehow GMail mistook my attempt to import some of my old emails into it for an attempt to forward them all to the original recipient. Apologies if you’ve gotten old (probably TermiSoc-related) emails from 2005 or so.

Add a comment