Mozilla 1.2 Betas Start Flowing 367
Asa Dotzler writes "Today mozilla.org released Mozilla 1.2alpha. This is a preview of what's to come with Mozilla 1.2 expected in early November. The new alpha contains great new features like Type Ahead Find which allows quick web page navigation when you type a succession of characters in the browser. In addition to the new features Mozilla 1.2a contains stability and perfomance improvements including a major boost in the speed of downloading mail on Mac OS X.This release comes on the heels of the security and bugfix follow-up to Mozilla 1.0. If you're a 1.0 user and you're not upgrading to Mozilla 1.1 or newer then you are strongly encouraged to get Mozilla 1.0.1 for security and stability fixes."
Well at this rate... (Score:2, Interesting)
Seriously beyond the commonplace protocol upgrades and reworks I think that IE 7.0 will end up being quite the hard sell for the typical Windows User. This may present an opportunity for Mozilla/Netscape to steal a bit of marketshare if things go right. This will happen anyway as AOL is planning to move their browser engine over to Moz (already been done for the MacOSX version I believe) and the Gecko AOL betas run quite well.
J
No major news, and still a memory hog (Score:2, Interesting)
And I wonder if they're ever going to do anything about the memory footprint. Together with Windows 2000's awful VM handling, I'm in swap city every time I copy a large file, having to wait more than 30 seconds for my Mozilla window to be swapped back in.
Download Manager with no restart functionality? (Score:4, Interesting)
I have to wonder what the rationale behind including a download manager with no scheduling or restart functionality is.
Oh well. I assume that this will come along eventually, just like everything else. The team has fixed both the bugs I submitted for 1.1a (table layout problems), so I will assume that they will eventually get around to this kind of functionality.
Mail *downloading* speed? (Score:0, Interesting)
What the hell is it doing that it needs to run that slow?
Re:No major news, and still a memory hog (Score:5, Interesting)
http://themes.mozdev.org/skins/pinball.html [mozdev.org]
That said, Moz can be quite the memory hog, especially on graphically intense pages. One of the big mistakes I see that can aggrivate this is the practice of tiling single-pixel graphics over a huge area. I'm not familiar with the gecko code, but I'm guessing that rather than rendering the tiled image once and keeping a handle for the resulting bitmap, Moz renders the image over and over again as it tiles and keeps a handle for each tile.
PHPBB sites are particularly bad about this, since the 'Sub Silver' theme uses several images that are about 5 pixels wide x 30 pixels tall. 150 pixels total. If you have to cover an area that is 1000 pixels wide, you need 200 repetitions of that 5 pixel wide image. If you repeat that area 25 times, and keep seperate instances of the image for each tile, you end up keeping the image in memory 5000 times.
Anyone more familiar with Gecko willing to comment on the actual mechanism of how it handles tiled images like this?
Type ahead find (Score:2, Interesting)
That is the problem with the behaviours of the mozilla interface widgets : they don't behave like any plateform.
Would it be too hard to make the widgets behave diffently depending on the plateform ? For example, when you click once in the address bar, all the text gets selected. That works on Windows, but not on the Mac, where the standard is to insert the bar cursor at the point where you clicked. The same for clicking in the scrolling bars : it only pages once, not repeatedly like on a Mac. The same for the dropdown menu (see the comparison of the windows drop down menu and the mac one by Bruce Tognazzini), etc etc.
I think people like visual inconsistency (themes, skins), but hate behavioural inconsistency.
Bayesian anti-spam filters (Score:3, Interesting)
Remember that Slashdot article [slashdot.org] on Paul Graham's method of spam blocking through Bayesian filters [paulgraham.com]?
In case not, the basic idea is that spam can be fairly reliably detected through statistical analysis of word choice. For instance, a message containing the word "GNU" probably isn't spam, while one containing "remove" might just be (but see the write-up [paulgraham.com] for more detail).
Anyhow, there's been a bug filed [mozilla.org] requesting Bayesian filtering for Mozilla. If you're interested in the feature, you may wish to vote for the bug [mozilla.org] (of course, you'll need a free Bugzilla account [mozilla.org] to vote).
VI syntax for searching (Score:5, Interesting)
Wonder if it supports ? for backwards searching, i for case insensitive...
Type ahead find (Score:5, Interesting)
However, I'm slightly concerned about the description [mozilla.org] of this feature. I gather this appeared in IE, and I fear that mozilla is more concerned with "parity" than with the most usable implementation. (Do you realize that when using the mouse wheel to change text size, going up makes the text smaller? Copied from IE. Won't fix. Bug 146491 [mozilla.org])
It appears to start searching as soon as you type a letter. This rules out all other possible uses for the letter characters. All of the most accessible keys on the keyboard "used up", just to avoid having to hit a command key to start searching in links. Even though you already have to hit a command key ("/") to search in the full text. If we want more keyboard functions, only punctuation keys (or key combinations) are available. For example, to seach for "foo" I can type "/foo", but to get the next hit, I have to do Ctrl-G, instead of something convenient like "n". This seems shortsighted.
Well, I'll have to try it before I can be sure of my criticism, but from what I understand, this feature could become much more powerful if the implementors design it well, instead of merely copying IE.
Re:Well at this rate... (Score:4, Interesting)
How about that goddamned modal dialog window [mozilla.org] that pops up when it can't load an unreachable embedded element.
Please don't whine [mozilla.org] about how it's not nice to alias whatever.doubleclick.com to 127.0.0.1 in my hosts file. I know it's a kludge, but it's my hosts file. I don't want any traffic to go to those domains, whether it's from Mozilla or any other application.
Bug 28586 has been open for over two years and has 115 votes against it. (Moz team, please just swallow your pride and deal with the fact that your users just might not use their machines the way you do.)
(And the fact that hosts-based blocking is a kludge doesn't change the fact that modal dialogs for "document contains no data" or "ain't no host there" are just plain evil. The domain serving an image might be Slashdotted, for instance.)
Until I switch to Mozilla for everything, I still need my hosts-based blocking for the crap my proxy doesn't catch.
Of course, if I keep having to click on its goddamn modal dialogs instead of just seeing "X"s or broken image icons when a site's images are Slashdotted or blocked by my hosts file or firewall, I'll never use Mozilla as a web browser, let alone switch other parts of my life over to it. Pity. Apart from this bug, it looked pretty cool. But with this bug, it's unusable.
This has to go into the main builds.
(Disclaimer: if this made it into the 1.1 release, I confess I never bothered checking. Anyone knwo if it made it into 1.2? I can apply the patch and build the damn binaries myself if I have to, but most Joe Sixpack users can't.)
Killer feature! (Score:2, Interesting)