4 ms·
Any data available on the # of crashes in the last month? FF4 is great in every way except the huge memory footprint and the fresh bugs.
by pdenya 16y ago
Any data available on the # of crashes in the last month? FF4 is great in every way except the huge memory footprint and the fresh bugs.
- magicalist 16y agocheck out https://crash-stats.mozilla.com/products/Firefox https://crash-stats.mozilla.com/products/Firefox
- Silhouette 16y ago> FF4 is great in every way except the huge memory footprint and the fresh bugs. And the typography engine that does kerning so badly that it renders several letter combinations in normally clear fonts literally illegible at smaller sizes used by many web sites. And the fact that they still haven't properly fixed Java applets. And the terrible usability issues. And worst of all, the fact that for all the cute gimmicks with tabs, it still runs everything in the same process, which is a fundamental robustness, security and performance problem that every other major browser fixed years ago. I write browser-based code for a living, and IMNSHO Firefox 4 is without question the worst of the major browsers on any major OS today as far as technology goes. Such redeeming features as it does have come almost entirely from third party extensions.
- ootachi 16y ago"And the typography engine that does kerning so badly that it renders several letter combinations in normally clear fonts literally illegible at smaller sizes used by many web sites." The native OS typography engine, you mean? Firefox uses the native text rendering stack on every platform it supports. "And the terrible usability issues." Like what?
- Silhouette 16y ago> The native OS typography engine, you mean? Firefox uses the native text rendering stack on every platform it supports. That's interesting, because every other browser I've ever used on my WinXP system (including Firefox up to 3.6) could display a 'T' followed by another letter without screwing up the positioning, as can every other piece of software I use on the system for that matter. I suspect you are mistaken, and Firefox 4 is actually using something other than vanilla Windows typography where it didn't before. > Like what [terrible usability issues]? Well, just off the top of my head... They messed up just about everything on the address/history/awesome/whatever bar: refresh is now a tiny button on the other side of the screen, home is even further over squashed between icons for extensions, you are supposed to realise (presumably through some divine revelation) that holding down the mouse button over the back icon is how you jump back by more than one page, there is a tiny, almost invisible arrow on the right of the tab line that shows all tabs and lets you open a Tab Groups page with a weird UI that doesn't fit in at all, etc. They moved "Bookmark all tabs" from an obvious place on the Bookmarks menu to the context menu when you right-click a single tab. (A Google for "Firefox 4" "bookmark all tabs" currently returns nearly 30,000 hits.) Go to Tools|Add-ons with nothing open and you have no obvious way back to a regular browser tab. Autocomplete of password fields on forms seems to be unreliable. I'll stop there, because I think the point is made well enough.
- capnrefsmmat 16y agoFirefox 4 still uses GDI on Windows XP for font rendering. I'm surprised, because the majority of the complaints came from its switch to DirectWrite on Vista and 7; DirectWrite still has display issues, though they're Microsoft's problem, not Mozilla's. Also, try right-clicking the back icon, rather than holding down the mouse button. And using the toolbar's Customize system to move Refresh and Stop and Home wherever you want them.
- Silhouette 16y ago> Firefox 4 still uses GDI on Windows XP for font rendering. It also now uses an OSS library, HarfBuzz, to position the glyphs. The kerning problem is well documented, e.g., see here: http://forums.mozillazine.org/viewtopic.php?f=39&t=2158847 http://forums.mozillazine.org/viewtopic.php?f=39&t=21588... > Also, try right-clicking the back icon, rather than holding down the mouse button. Sure, you can get the list either way, but why change from the widespread convention of having a down-arrow for a drop-down list as earlier versions did? The newer version is less discoverable, and making that sort of change without a good reason is just poor usability, pure and simple. > And using the toolbar's Customize system to move Refresh and Stop and Home wherever you want them. I hadn't even realised Firefox 4 still had such a Customize option until you mentioned it. To activate it, I have to navigate my mouse cursor to a region about 3-4px wide (on my 1920x1200 screen) between some of the areas on the bar, and right-click, even if the mouse cursor has already changed to the left-right arrow in some of those places. Once again, this is just poor usability (as is moving the buttons around in the first place when upgrading from an existing Firefox 3 installation that had them in a different place where they could still go).
- TwoBit 16y agoWhat you're complaining about is called "kerning pairs" and is considered to be superior text layout technique. Most professionally laid out text in print uses this. See http://en.wikipedia.org/wiki/Kerning http://en.wikipedia.org/wiki/Kerning. Perhaps you aren't used to it.
- 16y ago
- lambda 16y ago> every other major browser fixed years ago Chrome and IE are multi-process (though IE provides less protection as more tabs are included in the same process). I don't think that Webkit2, the multi-process Webkit that Safari will use, has been released yet. I believe it's intended to be released with Safari 6, whenever that happens. I don't believe that Opera is multi-process. So, out of the 5 major browsers, two are multi-process, one is in the process of being made multi-process, and two are not.
- Silhouette 16y agoNo offence to Opera, but I don't consider it a major browser when it has about 1-2% market share depending on who you ask. I confess to being wrong about Safari, though, if you consider its 5-9% enough to count: I first saw details of WebKit2 about a year ago, and hadn't realised that it's not going in until Safari 6, which in turn isn't expected until OS X Lion some time this summer. Nevertheless, I stand by my original point: IE, Firefox and Chrome between them represent by far the lion's share of the browser market, and IE and Chrome have both had a separate process model for some time now.
- lambda 15y agoI consider IE, Firefox, Chrome, Safari, and Opera all as major browsers. Sure, Opera only has 1-2% global market share; but consider the fact that 1-2% global market share means somewhere in the neighborhood of 20 million users (assuming estimates of 2 billion global internet users is correct). I consider that a fairly significant browser. Also, there are markets in Europe where that percentage is much, much higher. And those European markets are generally a lot more relevant to me than China (and to an extent other Asian countries like South Korea), where a lot of the IE usage comes from. Due to the greater internationalization, cultural, and regulatory barriers in China, it's not a market that interests me all that much. I generally consider the major browsers to be ones which have their own rendering engines (though Chrome and Safari share one, they both do significant development of that engine and each have their own JavaScript interpreters), actively attempt to be compatible with the entire public web, participate in the web standards process, introduce innovations to the web platform which are later adopted by other browsers, and have a significant number of worldwide users or users in major markets. IE, Firefox, Chrome, Safari, and Opera all meet those criteria.
- rpearl 16y agoJust a bit of commentary... Note that it's not "one process per tab". Chrome has "process-per-site", and "process-per-tab" (where a "tab" is not one tab, but a bunch of related tabs.) https://sites.google.com/a/chromium.org/dev/developers/design-documents/process-models https://sites.google.com/a/chromium.org/dev/developers/desig... In general, IPC is slower than threading--Every time you switch processes you have to flush the TLB, as opposed to switching threads, which only has context switching overhead.