Chrome Release Cycle(docs.google.com)
docs.google.com
Chrome Release Cycle
https://docs.google.com/present/view?id=dg63dpc6_4d7vkk6ch&pli=1
6 comments
Lack of silent updates was one of the key reasons I switched from Firefox to Chrome (speed was another.) Seemed like every time I opened the browser to view a webpage I needed to see that very moment, Firefox would pop up an annoying update prompt that I'd keep clicking out of each time.
+1 - Interesting - that's actually why I've switched over to Chrome as well. Basically, every time I started up Firefox on my System at work on XP, Firefox would Freeze with a "Hey, we're going to upgrade you now - why don't you just go grab some coffee!" - So, slowly but surely, my goto browser started to become Chrome. The downside of this, is that, over time, I started firefox even less frequently, and, as a result, it was _always_ waiting for an upgrade. making it even worse.
Chrome just starts for me. I'm happy with it.
Chrome just starts for me. I'm happy with it.
The funny thing is that from a Linux user's perspective, Google was simply providing a workaround for what should have been the OS's job. On Linux, Firefox's auto-updates are disabled so that the OS can handle that instead.
This is probably the best solution, as it gives you the most control over updates, and applies to all software system-wide. You're not forced to install updates at any time, and you can disable the notifications if you want.
This is probably the best solution, as it gives you the most control over updates, and applies to all software system-wide. You're not forced to install updates at any time, and you can disable the notifications if you want.
The update popup for extensions is still annoying though - even on linux.
Like I said, you can disable that.
Many distributions' version of Firefox comes with that disabled out of the box.
And last I looked, that's how Chrome/Linux works too, at least on Ubuntu. When you download and install the deb, it'll set up apt sources so you get updates through the system.
Hah! At the moment I read "at that very moment", I got a popup from Chrome's PDF reader in the background tab I loaded the article in:
"Some slides have failed to load and may appear blank. Click OK to try reloading the presentation, or cancel to continue viewing."
So Chrome hasn't completely ridded itself of modal popups just yet, though it's happens infrequently enough that I was pretty surprised.
"Some slides have failed to load and may appear blank. Click OK to try reloading the presentation, or cancel to continue viewing."
So Chrome hasn't completely ridded itself of modal popups just yet, though it's happens infrequently enough that I was pretty surprised.
I also appreciate that when I close a tab, its memory is returned to the OS. My safari process tends to need killing once in a while, after it's been mean to my VM.
I always wonder why those update dialogs pop up when you start a program. That’s the one point in time when it is very clear that you definitely do want to use the program, the one point in time when you don’t want to be disturbed. Those dialogs would be a lot less annoying if they popped up when you quit the app, still allowing you to opt-out of updates but in a lot less annoying way.
One possible negative effect is that updates get installed later (an average session length later), I wonder whether that’s offset by people not installing updates when the browser starts because they have something better to do?
One possible negative effect is that updates get installed later (an average session length later), I wonder whether that’s offset by people not installing updates when the browser starts because they have something better to do?
My absolute favorite in terms of updating is Calibre; an otherwise excellent app (albeit with a fairly "experimental" UI).
Calibre tells me about a new release every single time I open it, they seem to release almost daily.
Sadly the "update" button on that dialog merely opens the web-browser, where you have to manually download and install the new version. Oh, and that is a 60MB(!) download every time.
The procedure is so absurd that it's actually amusing again. That's why I'm still doing it, I guess, even though after 20 or 30 updates I still haven't noticed any visible change to the software.
Calibre tells me about a new release every single time I open it, they seem to release almost daily.
Sadly the "update" button on that dialog merely opens the web-browser, where you have to manually download and install the new version. Oh, and that is a 60MB(!) download every time.
The procedure is so absurd that it's actually amusing again. That's why I'm still doing it, I guess, even though after 20 or 30 updates I still haven't noticed any visible change to the software.
That has happen to me sometimes with Chrome and every single time I think how cool it is that it updates itself and how cool it would be to just keep the bug free version around.
I rely heavily on the information saved by Chrome every time I use a form and I noticed how it doesn't save it when I use the beta version. Now that's fine because I can just use the last non-beta release just like everybody else but on the current one (at least on Windows 7) I can't access those saved values if the form is loaded by ajax into some iframe. It's really annoying.
I rely heavily on the information saved by Chrome every time I use a form and I noticed how it doesn't save it when I use the beta version. Now that's fine because I can just use the last non-beta release just like everybody else but on the current one (at least on Windows 7) I can't access those saved values if the form is loaded by ajax into some iframe. It's really annoying.
It's not only less of a hassle, but veritably good for the web.
If only we could essentially rely on all our users' browsers being up-to-date!
If only we could essentially rely on all our users' browsers being up-to-date!
One thing that helps Chrome release often is that the api for building extensions is all javascript. By comparison, Firefox extensions can have binary components. This means there needs to be a certain amount of platform stability or extensions will break every six weeks. I think that in the future we'll see faster release cycles for Firefox as the api for extension building becomes more js based and less dependent on xpcom components.
I don't know about that. I would love to have an extension that made the omnibar a lot more like firefox's awesomebar (good indexing, can search for key phrases for title of article and the such) but to my knowledge that's pretty much impossible because of the javascript limitation (I hope I'm wrong here, not really sure).
I just wish google was a bit more subtle about wishing people search on google again when looking for previously viewed sites.
I just wish google was a bit more subtle about wishing people search on google again when looking for previously viewed sites.
Funny, I wish Firefox's awesome bar let me press enter to go to the first result without pressing down first.
Almost every extension I've tried has worked fine on Firefox 4, and certainly between the 3.x releases. While you might be right that this should be considered occasionally, it doesn't seem to be a big deal.
Firebug, HTTP Live Headers, AdBlock Plus and a slew of other extensions I typically use do not work until a late beta is out.
Chrome extensions can have binary componants with NPAPI : http://code.google.com/chrome/extensions/npapi.html
One 'problem' with this modus operandi is that some companies use the big releases as selling points. It is interesting that the Chrome team does not feel the pressure to do this. Especially considering how heated the browser war is these days. Everyone seems to think Firefox 4 is going to be the salvation, for example.
On the other hand, the silent updates are a blessing, and good features are another way of attracting users.
On the other hand, the silent updates are a blessing, and good features are another way of attracting users.
As the presentation explains: market features, not versions.
Well, Google does make some noise about releases. I forget which version (proving your point) of Chrome brought the Store, but at the time it was touted by version number, despite the silent update. Probably something we'll see more of as companies move to the cloud and mobile.
The version number was touted by tech blogs and other media, but not by the Chrome team themselves, a point which they made in response to a comment here on HN if I remember correctly.
I like the idea of hiding features behind flags. You can see evidence of this from the number of command line options in the source:
http://src.chromium.org/svn/trunk/src/chrome/common/chrome_s...
They do this even for fundamental features of the browser. For example, there's four different process models available:
http://dev.chromium.org/developers/design-documents/process-...
For planning software releases, I take inspiration from Chrome, Ubuntu, and Microsoft. They all have different release cycles, it's interesting to compare them.
Chrome: agile, 12 weekly releases, subtle upgrades, parallel versions with different stability levels.
Ubuntu: fixed six monthly schedule, well advertised schedule (e.g. http://www.youtube.com/watch?v=107E4RXNxYc) means difficult for them not to release on time, it's a selling point. Every 24 months long term support release.
Microsoft: major products released approximately every three years. Heavy marketing efforts around releases rather than features. Service pack releases and patches released on an ongoing basis.
I guess Chrome has less of an enterprise install base, which means they are more able to release on an ongoing basis. Enterprise IT teams like to install known good releases and not let other the vendors manage the upgrades.
http://src.chromium.org/svn/trunk/src/chrome/common/chrome_s...
They do this even for fundamental features of the browser. For example, there's four different process models available:
http://dev.chromium.org/developers/design-documents/process-...
For planning software releases, I take inspiration from Chrome, Ubuntu, and Microsoft. They all have different release cycles, it's interesting to compare them.
Chrome: agile, 12 weekly releases, subtle upgrades, parallel versions with different stability levels.
Ubuntu: fixed six monthly schedule, well advertised schedule (e.g. http://www.youtube.com/watch?v=107E4RXNxYc) means difficult for them not to release on time, it's a selling point. Every 24 months long term support release.
Microsoft: major products released approximately every three years. Heavy marketing efforts around releases rather than features. Service pack releases and patches released on an ongoing basis.
I guess Chrome has less of an enterprise install base, which means they are more able to release on an ongoing basis. Enterprise IT teams like to install known good releases and not let other the vendors manage the upgrades.
Also: Microsoft is charging for updates. I'm not sure people are ready for a paid subscription service for windows.
Indeed. Microsoft charges for their major version updates. I'm sure this is a reason why people have put excessive importance in Chrome's major version number. When you release for free you can choose version numbers for engineering reasons, when your revenue is tied to releases you ensure that each major version bump is sufficiently "meaty" to justify the purchase.
"Market features, not versions," but to be sure this is Chrome's most important feature. It also mirrors what's happening in the development of HTML. http://www.mail-archive.com/[email protected]/msg23160...
If IE had been engineered in the same way, we never would have gotten into the IE 6 trap and when all browser manufacturers adopt this strategy, web technology will move much more quickly.
If IE had been engineered in the same way, we never would have gotten into the IE 6 trap and when all browser manufacturers adopt this strategy, web technology will move much more quickly.
This came out at just the right time. I've been pushing my team towards a releaSe cycle very similar to this one, not modeled after any particular system, just seemed like a good. Idea. Nice to have some outside vindication from such a respected source.
I'm sure I'd be singing a different tune if it broke my browser during a silent update.