FireFox 1.5

Everybody seems to be writing about the new firefox and as a loyal user, I shouldn’t stay behind. I’ve been using the alpha and beta versions and they have proven to be mostly trouble free (as much as you could expect from alphas and betas).

The alphas were a little unstable and I encountered a couple of crashes (over a period of several weeks). That hasn’t happened to me with the beta which I’ve been using since it was released. The largest problem has been the intially slow pace with which extensions are tested and ported. By now most of the more relevant (and actively maintained) ones have been ported. There are still a few extensions missing. Presumably the developers are waiting for the final release or simply not interested in immediately fixing their extensions.

Now the new features. At a first glance, not much has changed. Most changes are under the hood. The most visible change is no doubt the rearranging of the preferences dialog. IMHO it doesn’t make much difference but I agree it is slightly nicer this way. Also the rss button has relocated from the status bar to the locationdropdown. IMHO this is slightly awkward or at least unconventional. And finally FireFox now has error pages for timeouts and other server errors, just like IE has had for years. Other than that, I can’t think of any relevant, eye catching changes.

Then there are the changes under the hood. SVG and canvas are two new cool technologies. However, since internet explorer does not support these, I doubt they will be used very much. Firefox only sites are a bad idea IMHO. Sadly, this also applies to the CSS improvements (e.g. css 3 columns) and javascript improvements. Nice, but it’s not likely these technology improvements will see much adoption (except maybe in browser extensions where browser compatibility is not an issue).

The most visible change to end users will be the performance increase. When 1.0 was released last year, it was released off a branch which had been frozen quite some time by then. Meanwhile development on trunk (a.k.a. mozilla 1.8) continued and about 1.5 year of development has accumulated there, most of which will end up in FireFox 1.5. FireFox 1.5 has a new clever mechanism for back and forward which prevents it from rerendering a page and caches the results. Consequently, going back and forth between pages is now fast. Also normal rendering of pages has seen some improvements. It is hard to quantify this but it certainly feels faster.

Something that is often overlooked in reviews such as this is extensions. As I said there are some porting issues but these will go away when the final release approaches. What is worth noting is that some extensions have really evolved from nice toys to real valuable browser additions. IMHO the added value these extensions provide is often overlooked. You can’t really imagine doing web development without such powerful tools as the javascript debugger, the webdevelopment toolbar, live http headers and many more developer oriented tools. As a mere user, adblock, sage, forecastfox and context highlight make life so much easier and fun. And there are loads more of mature, well tested, valuable extensions.

So to summarize, some minor feature and gui development, lots of under the hood changes which you will notice in the form of snappier performance. Other than that it is still FireFox, a damn good browser, a solid upgrade from any browser you currently use.

umlet

After years of looking for a UML tool that doesn’t suck I stumbled upon Umlet: http://qse.ifs.tuwien.ac.at/~auer/umlet/. Umlet is a very simple tool that makes it very easy to do what IMHO is the primary thing that software designers look for: quickly sketch some diagrams without worrying about correctness, completeness and details. When I’m in design mode, details distract me and break my concentration. I use a design tool primarily to structure my ideas and secondarily to illustrate them to others.

Umlet is very nice in that it doesn’t try to enforce anything. You just drag uml shapes to the drawing area, specify as much (or as little) details as you need in a brilliantly simple text area. This is a killer feature that I have never seen anywhere else. Rather than navigating through complicated dialogs you just type
a

attribute

operationFoo(bladiebla)

to specify the contents of a class rectangle. Other than the dashes there are no restrictions. Don’t want to specify attributes: leave them out. Don’t want to specify the parameters: then don’t. Whatever you type appears in the appropriate place.

When you are done you save it as a bitmap, pdf or svg drawing and it is ready to be pasted in your design doc or powerpoint presentation.

Sounds simple. It is but try to work like this in rational rose, visio or togetherJ and you will quickly get frustrated by these tools insistence to only allow semantically and syntactically correct diagrams. These tools have a purpose in an ideal model driven world that IMHO only exists in the minds of over payed consultants working for companies that are foolish and rich enough to pay the rediculous license fees these tools requires. Outside this world they are bloated, unusable tools that mostly lie unused on the shelves of software architects who have better things to do.

Umlet beats these tools easily if all you want is draw some free form uml diagrams and it comes with an unbeatable price. Of course it isn’t perfect. It doens’t generate code, the user interface is primitive, the set of export formats is limited, etc. But it gets the job done when it comes to sketching designs quickly and most importantly it is light weight and does not impose restrictions, development methods or any particular design solutions.

FireFox Alpha a.k.a Deerpark

Quite uncharacteristically, I have not been touching any nightly builds of firefox since 1.0. Part of the reason is that the mozilla developers seem to have abandoned the notion of bi-monthly milestones (so it has been a long time since any reasonably stable build). But today I gave the DeerPark alpha rc1 a try. That means it is a first release candidate of a first alpha of what will be Mozilla FireFox 1.1 some day. It is now named Deerpark so regular Firefox users don’t touch it. Probably this is a good thing because you can expect things to break down when using alpha software.

Deerpark is a nice browser. Of course it has a few little quirks (hey it’s an alpha build) but you can browse with it and overall it is as pleasant to use as its predecessor. Deerpark is not revolutionary in its interface. A few minor tweaks in the user interface are all you will notice at a first glance. The most significant change is in the options pane where you now have a horizontal icon bar instead of a vertical one. Not an improvement IMHO but I rarely use it anyway. Also some of the preferences have been rearranged but nothing revolutionary here too.

Under the hood the biggest change is svg rendering. Svg rendering has been under development from about 2000. I recall trying svg builds years ago. Svg is one of those more or less failed w3c standards that still await widespread adoption (other than niche grahphical products and linux desktop decoration). Deerpark could be what triggers this adoption.

Another notable but mostly invisble change is gecko that is now a year older than the version shipped with FireFox 1.0 and presumably has had quite a few tweaks (performance, standards compliance, rendering bugs, etc). Apparently they also fixed inline editing.

Other than that the fixes are minor. Deerpark is a nice incremental change but nothing revolutionary.

So why am I back to FireFox 1.0.4? Answer: extensions. I need my extensions. In particular sage (http://sage.mozdev.org) is important for me and sage needs to be fixed for deerpark. Several other extensions I use, also need to be fixed. I suspect many extensions will be fixed in the next few months. Probably the Firefox/deerpark beta will be a moment when many 1.0 users start to switch.