September 20, 2002

Missing the Point

A lot of people seem to be missing the point of Phoenix, as evidenced by the responses on Mozilla News and MozillaZine. Let me emphasize something here: if you think Mozilla's current UI is acceptable, then you are clearly not the target audience for Phoenix.

Here is a quiz to test whether or not Phoenix is the right browser for you.

  1. The link toolbar is:
    • critical to my day-to-day use
    • vital when using Bugzilla! Doesn't everybody use Bugzilla?
    • link what?

  2. The sidebar is
    • indispensable since i run at 1600x1200 resolution
    • not cool enough, since i can't float and dock all my panels and have splitters between all panels and see web page progress for individual HTML panels and check my email entirely from within sidebar and...
    • a waste of real estate

  3. Form auto-fill
    • is not useful for me unless I can fill out 20 pages of personal information first
    • should just happen automatically

  4. Downloads
    • should take place in a tree view with progress meters in the columns!
    • should be clearly visible and understandable.

  5. Toolbars should
    • be dockable to all four corners of the screen, be able to float outside the window, be fully customizable such that I can make my own custom commands, be able to edit existing buttons' commands, be able to create my own toolbars, be able to put toolbars on the same line, and be able to edit the submenus and context menus of items (including the items on the menu bar) and browse my file system and cook me dinner and wash my car and walk my dog and do my taxes and mow my lawn and...
    • be customizable within reason

  6. Composer should
    • always come with my browser. I want composer options all over my UI. Everywhere!
    • not be part of a Web browser.

  7. Mail should
    • be part of my browser program. Aren't they the same app? There is a difference?
    • be a separate application.

Now to those people who want the full-blown functionality of the Mozilla trunk, you can still get that with Phoenix. The idea is to relegate more features to the "optional add-on" category. If you want the link toolbar or the sidebar or mouse gestures or any other features, you can download and install them yourself. I expect Phoenix will have a little add-in manager that will facilitate this process. There is currently an expectation on the part of an alarming number of people that every feature implemented by anyone should automatically be part of the default Mozilla install/download.

Why?

A layered approach scales better. You can then have a browser that can become as complex as you want to make it, but the choice is left in your hands. The geek features aren't inflicted on you by default.

Finally, remember that Phoenix's UI is not controlled by Netscape. This is an opportunity for some of the core Mozilla Navigator developers to build the browser that they have always wanted to build, without having to compromise the user interface to satisfy the various conflicting pressures exerted by factions within Netscape.

Posted by hyatt at 3:23 AM | Comments (2)

September 12, 2002

A Line in the Sand

Bug 22056 has to do with enabling different toolbar modes. It's a pretty basic browser feature that has been missing from Navigator for years. Even simplified browsers like Chimera have this feature. Neil did some excellent work in 22056 and his code finally landed. It naturally spiked startup time and window time slightly, and so it ended up being backed out because of Mozilla's no tolerance policy for regressions.

While this "line in the sand" policy is in many ways a good one, I feel like it misses the point. There is a natural tendency when designing applications to add features in every new version of the software. Only rarely do you see features removed. With each passing version, you get more and more bloated, relying on faster machines and more memory to save the day. Who cares if the user interface is now full of 3000 menu items? You still support every last feature since version 1.0, so no customers can possibly be dissatisfied!

You can really only cram so many features into a product before its size requirements and performance requirements have to change. This is an obvious rule. It's like you start with an empty elevator that says "Capacity: 10 people." The elevator stops at the first floor (version 1.0) and a bunch of people (features) get on. Continuing on its journey, the elevator stops at the second floor, and still more people get on. The elevator is now full, and it continues to the third floor (version 3.0). Unfortunately the elevator is full, but there are a bunch of people waiting at the third floor to get on. Some of them squeeze in anyway, past the fat person from the first floor (the Mozilla sidebar feature) who is taking up enough space for 3 people. Everyone wishes he'd get off at the third floor or even the fourth floor, but he doesn't. Someone (the Mozilla wallet feature) from the second floor cuts one on the way to the third floor, so he's useless, and everyone wishes he'd get off too. He doesn't though. Nobody does. People keep piling in at floor after floor, until eventually the elevator support cable snaps and everybody dies.

We need to forcibly eject people from the elevator. Remove the features that nobody wants and replace them with the features that matter. Cull out the features that didn't work in Mozilla 1.0 and make sure they aren't there in Mozilla 2.0. Make more of the features optional plugins so that geeks who want some of the more obscure features (and that have powerhouse machines) can go download them on their own. Only if we actively fight the trend towards bloat will we finally produce an awesome Mozilla browser.

Posted by hyatt at 1:16 PM

Priceless

I just finished reading an article about Mozilla for Salon.com. This excerpt was rather amusing.

It is a good question, because in almost every way, Mozilla is a better browser than Navigator. For example, Netscape's best new feature, tabbed browsing -- which lets you have several Web pages open in the same browser window, and allows you to bookmark all the pages under one name -- was in Mozilla many months ago, and the Mozilla project that created it (called MultiZilla) already has an improved version available. When asked about this, Yecies, of Netscape, said, "That's true, but the engineer who's working on it [for Mozilla] is a Netscape employee. It was always done with the intention of fostering general browsing usability for Netscape."

Yes, ok, I suppose that's true if by "Netscape employee" you really meant "Apple employee." and by "always done with the intention of fostering general browsing usability for Netscape" you meant "was done in a weekend for Mozilla because I thought MultiZilla was cool."

Here's how the whole tabbed browsing thing happened. One night I finally downloaded an extension called MultiZilla (go check it out on mozdev.org. I was particularly impressed with a feature contained in MultiZilla called tabbed browsing. I started doing research and discovered NetCaptor, a tabbed browser that embedded WinIE.

MultiZilla was cool but at the time suffered from two fundamental flaws that prevented the code from being incorporated into the Mozilla tree. The first was a UI flaw, namely that at the time it had ripped off NetCaptor down to the last context menu item. The GUI was similar enough that there would have been definite concerns about so obviously copying some of NetCaptor's more obscure capabilities (like sticky names and tab locking). The second concern was that the tab behavior wasn't encapsulated cleanly into a widget.

I produced a simplified version of tabbed browsing on my own time (did it in a weekend) that removed some of the geekier NetCaptor features and that encapsulated the tab behavior so that the changes to other Navigator files would be minimal. Once I established that it didn't degrade performance in the single tab case, I checked it in as an experiment.

The response was overwhelming, as were the bugs that started being filed. So much so that at first I wanted to back tabbed browsing out of the tree. I was overruled by Mozilla, which turned out to be a good thing for all I think. :) Even with all the excitement and hoopla surrounding the advent of tabbed browsing on the engineering side (and in the Mozilla community), Netscape still didn't get it. Netscape marketing prioritized all sorts of useless work that nobody had even started above tabbed browsing in their marketing document. They continued to do so for months, simply not getting it. It was this odd curiosity that one of their engineers had checked in, and they didn't know what to make of it.

Only after the press raved about it did Netscape really jump on board. I'm sure Netscape is doing the same thing now with popup blocking. Can't you just see it now? We'll have a Popup Manager, and a Manager to manage the Popup Manager, and twenty-seven preferences for fine-grained control of all aspects of popups.

Can you believe how disfunctional Netscape is? When their engineers say "you should do this" or "you should do that", they get completely ignored (or blown off), but when CNet says "We didn't like this, or we didn't like that.", Netscape scurries to meet their demands. That is simply pathetic.

Posted by hyatt at 12:50 PM

September 3, 2002

Usability Problems with Mozilla

Blake blogs about how mpt wants Mozilla to look just like MSIE. I have to admit, the evidence is pretty compelling. I recall someone asking me, "Do you really agree with mpt's Top 10 list? He's quoted you at the top of the list!"

Do I agree that those ten items mpt mentions are the top ten problems? Of course not. No two people will have the same top ten problems. Also keep in mind that mpt and I can agree that something is a problem without necessarily agreeing on the solution to the problem. Maybe we have different ideas regarding how to solve a particular issue, but we at least both believe it is an issue that needs to be addressed. That's something.

To cover the list specifically:

  • Navigator chrome structure - While I don't necessarily agree with mpt's proposed default configuration, I do agree that the chrome structure is painfully restrictive, and that customizable toolbars need to be implemented in order for us to acquire the flexibility to deal with this problem.
  • Speed - can't argue with this, except to say that cutting out a lot of the useless UI and features from the chrome helps substantially. Reduce bloat, gain speed.
  • Text editing - if you use Chimera on the Mac, you'll see that the textfield widget is easily the most painful part of the entire application. It's buggy, slow, misbehaves, and doesn't edit the way you'd expect. This is IMO Chimera's top usability problem.
  • Message display - Yes. No argument here.
  • Search - Yeah, it's a mess. Don't know if it would be in my top ten, but it's a mess.
  • Menu structure - This gets back to my blog about how the apps should be separated. The menu structure has been complicated in order to deal with multiple applications. A clean separation naturally simplifies the menus (e.g., you can eliminate the New submenu easily).
  • Migration - A problem, but IMO not one of the top ten facing Mozilla.
  • Context menus - mpt complains about two-click context menus, and yet, the OS default on Win32 (overwhelmingly) is to bring up a context menu on a mouse up. If you don't like it, complain about Win32, but don't cite this as a Mozilla usability problem when we're following the conventions of the operating system. (We simply listen for the WM_CONTEXTMENU message. That message fires when Win32 wants to fire it.)
  • Validation - Err, no. Not a usability problem. To the average bear, this is completely irrelevant.
  • Preferences - IMO this should be much higher on the list. Preferences are a tangled pathetic mess. Again, separating prefs for individual apps into unique dialogs would simplify things a great deal, but we should also remove nearly half the preferences that exist from the GUI. Mozilla is ridiculously overconfigurable.

Posted by hyatt at 11:30 AM