Lets not get upset and start to fight over some silly C++ bashing blog post. I checked out the C++ ISO standard draft and the result from Callum's example is undefined. Look at chapter 14.7.1, paragraph 12. For those C++ lovers out there who prefer GTK+/Gnome over Qt/KDE (how can they? ;-) there is the gtkmm bindings. That is, Callum's invalid code can be a part of a Gnome app as well as a KDE app, so lets not turn this into a flame-war.
I've been somewhat productive today. First I've been merging corrections into the Independent Qt Tutorial. Thanks for all the feedback. Then I've been splitting the SpeedCrunch site into a translatable part and the english-only parts (download and development).
Next in the pipe-line is the first release candidate for SpeedCrunch version 0.7.
I've been really busy the last couple of weeks but now I'm going on a business trip for the coming two weeks. That means lots of boring nights to kill of at the hotel (I'm not a bore, it is just that I'm going to Jokkmokk, so the nightlife is limited), i.e. hopefully a new release of SpeedCrunch and an update of the Independent Qt Tutorial - Qt 4 Edition (TIQT-4 from now on).
After a far too long lack of updates, the Independent Qt Tutorial is finally joined by a Qt 4 Edition. This time things are going to be done the Right Way, so everyting is scripted from a single source. The source code examples are copied directly from the compilable example sources into the HTML document and all off-line contents will be generated automatically. I write will be as there is alot of work left:
The chapters currently in the pipe-line include a quick presentation of the available widgets with examples, an in-depth look at the model/view concept and then a look at custom widgets.
But lets focus on what is available: an introduction to Qt and the signals/slots concept and then a run through a Qt application being built using Desinger. All chapters link directly to Trolltech's documentation when appropriate and all code is compilable. Check it out and tell me what you think at e8johan -at- gmail -dot- com.
At Chalmers, Human-Computer Interaction was not my favorite course. I thought that it was a too soft subject and nothing for real engineers. Now, I've come about and changed my mind.
The user interface of SpeedCrunch can be stripped down to the bare essentials. This saves screen estate, but does not show the unexpecting user that this is a calculator.
To show the user that there is more to it, the Evaluate button is available, and a clear all button is thrown in for good measure. These buttons are optional. All screenshots of this entry shows the same application with different settings.
This interface makes it clear that you can evaluate something and most users will start to type away at the keyboard and then press evaluate - or enter. However, the application does not look like a calculator and it does not tell the user what to expect to be evaluated. Hence, a key pad is available in the 0.7 version.
The screenshot above shows a different key pad layout than used in the 0.7 version. This keypad offers more common functions (asin/acos/atan instead of sinh/cosh/tanh - my maths tutor, Bernhard, will be disappointed, he really liked the hyperbolic functions). It also divides the keys into the groups digits and dots, basic counting, variables and advanced functions. Also, the "=" button has changed from assignment to evaluation since even I did use it in the wrong way. Now, the "x=" button shows how to achieve assignment.
Not to the interesting question. When offering all this user interface parts, how do I make the user turn them off to fully apprechiate the purpose of SpeedCrunch? There is the Windows way with balloons asking the user this now and then but that just makes me annoyed. Then there is the start-up tips, but I mostly ignore those and turn them of at once.
The only solution that I can come up with that is intuit enough would be to offer a shrink button (an arrow or a cross) that would allow the user to do this. But in that case - should that widget go away with the key pad, or stay there to let the user get it back in the same way that it was hidden? A user may hide the keypad by mistake and then not find a way to get it back... at the same time, such a widget would take screen space from the advanced user.
I have just released a new snapshot of SpeedCrunch - this time with make install working...
Due to work related issues the SpeedCrunch development has not been as quick as I was hoping for - but now I've finally come about and released a new snapshot. Get it from the site.
The first beta release has been a good school for me. I'm thrilled about the feedback, but also surprised about the amount of questions to reply and oddities to investigate. To handle this I've started a Google Group for SpeedCrunch. I'm surprised how smooth it was to setup a mailing list and get it running.
The future of SpeedCrunch can be divided into two parts: the user interface and the features, so my plan is to have the UI polished for 0.8 and then add more features in 0.9. For example, support for binary, octal and hexadecimal numbers, boolean operators, the mod and div operators, support for high accuracy numbers (3.{3} = 3.3333333....). I'm more than happy to discuss the possabilities so drop a message if you feel that you have an idea!
I had a look at my web-stats today and saw that there has been 438 downloads of the windows installer for SpeedCrunch. During the same period there has been 1509 downloads of tar.gz-files, but that includes all other such files - so I really have no reliable stats for that (yet). Anyway - this start is far better than I ever expected!
Finally I've found the time to move and update the site for SpeedCrunch. It now lives at speedcrunch.digitalfanatics.org. Update your bookmarks!
Also, thanks everybody who has mailed me and gotten back to me. I never expected this kind of feedback!
Released: SpeedCrunch 0.7 beta 1
The next version of the revolutionizing
SpeedCrunch
has been
released. The new version is 0.7 beta 1 and is based on 0.6 beta 2.
The biggest changes are the addition of
a keypad
(can be hidden by
owners of small monitors/users of many windows), that the application
is using Qt 4 and that it is available for not only X11 but also Win32
(anybody willing and able to do Mac OS X builds are welcome to contact me!)
As the homepage hasn't been updated yet the only place to get hold of this release (until I update the homepage) is to download a source code tarball (md5sum: 494752554690b8271d42f7c07e6fbb00) or this Windows installer package (md5sum: 70aec1cbaf7f7b0fed11227d176c68b7). Get 'em while they're hot!
I was told that the release had to be done before Jan 19 to get the application into the next Kubuntu release, so I'm hoping for that - it would be cool. So, to all translators and Jean-Remy Falleri (manpage author), sorry for being so quick. I've dropped you all a mail when you are reading this!
New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.
Keep up with the latest Advogato features by reading the Advogato status blog.
If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!
