libgcal 0.9.6 plus akonadi resources 1.2

Dear friends

It is being almost 3 months since the last public release of both the library and the resources. In the library, no new features on this release, but some improvements and bug fixes. The laundry list goes:

  • restored ABI (as noted by Fathi Boudra, gcal_contact_get_im was missing after the work in supporting multiple im fields)
  • CMake and autotools buildsystem will install the library in INCLUDE_DIR/libgcal (as per request of Mario Ceresa so it will adhere to Fedora Core packaging policies). Autotools patch by Ademar Reis.
  • And a new cleanup function gcal_final_cleanup() to avoid a potentially crash scenario if libgcal is linked with something else that *also* uses libxml. Nice patch by Chris Frey (thanks a lot dude!).

But for the resources, some nice new features:

  • IM fields are supported (as a matter of fact, this feature was available in libgcal since 0.9.5 release, but it never made into the akonadi contacts resource). Patch by Holger Kral.
  • Fix a scenario were authentication would fail if both resources (gcalendar + contacts) are set while using different google accounts. Now user information is saved in KWallet per resource (and the key also relies on the username), this makes possible the next new feature.
  • Multiple users are supported as long you have 1 resource per user. Previously, that would not be possible, because I was saving a single entry in KWallet (so it was overwritten after each new resource inclusion). I have tested with 3 different users *at same time* each one with both resources (contacts and calendar, total 6 resources instances) and its feeling quite stable.
  • Fix in authentication dialog, now the configure message reads “Your Google mail username (or user@whaveter.domain.com)”. I received some many bug reports about this, mostly because the previous message didn’t clue the user that hosted accounts are also supported. I’m just not sure if the translation strings were updated…

So, if you are curious, download it from the libgcal project website and have a try (compiling instructions are available in the tarballs). Or wait a couple of months and it should be pre-packaged for you in your distribution.

Known issues:

  • Only the main calendar is synced… supporting multiple calendars within the same account is in my to-do list.
  • Please, only 1 resource type X 1 user account. I haven’t tested the non sense case of adding more than 1 resource instance type per user account (so you are on your own if you do this).

September 8, 2010 at 12:10 am 8 comments

Sliderlayout

It all started with a simple problem: I want to change from one list of things to another, doing a ‘slider’ like transition (like you would do for example in a music player when you choose an artist and go to another list with the tracks).

That sounds pretty simple, right? Just fire a QPropertyAnimation (or if you are using Plasma, a SliderAnimation is even better) and animate the position of 2 widgets.

But what if I want to have those widgets inside of a layout (e.g. QGraphicsLinearLayout)? After all, I want my UI to scale between different window sizes, screen resolutions (and in mobile devices, adjust to screen rotations).

Hum… you could draw your widgets *over* the linear layout and check for resize events of the main widget *while* correctly positioning your set of widgets in the scene. But that sounds a bit too much hack-ish.

I asked 2 friends about this very same issue and one suggested me to create a layout with this behavior. Initially, I was not sure about it, after all, I have never worked before with QGraphicsLayouts and I got almost no time to have something working.

Well, the good news are that it only took me a couple of hours to get this:

[blip.tv ?posts_id=3856759&dest=-1]

Explanation: it is a layout were you can add several widgets and move from one to another by calling next()/previous(). And since it is just a layout, can be inserted in another layouts (in this case, inside of QGraphicsLinearLayout).

Maybe a bit of code would help to explain… say that you have 2 widgets and you want to move next when the current one is clicked:

Item obj1, obj2;

SliderLayout *slider = new SliderLayout;

slider->addWidget(&obj1);

slider->addWidget(&obj2);

connect(&obj1, SIGNAL(clicked()), slider, SLOT(next()));

slider->setCurrentWidgetIndex(0);

and you are done.

The secret here is to keep hidden all the widgets (except by the current one) and in the slider event, animate the position of both widgets (the current and soon-to-be-current widget) while correctly repositioning the widgets in the scene (and setting to show state both widgets of course).

Since I have the movement methods on this layout as slots, I have to derive from both QObject and QGraphicsLayout (and that I’m not completely happy, anyone has suggestions to workaround that?).

I have tested this layout (integrated with *far* more complex widgets) in varied devices (from ARM@430Mhz up to Intel Atom@1.6Ghz))and it behaved pretty well, basically because I’m only animating the position of 2 widgets. It might happen that if you have an overly complex widget, that your hardware is not able to do the animation with good frame rate (but here you can workaround by hiding some of the content of this overly complex widget, and that will be probably the subject of my next post).

There are 2 other remarks about the code:

  • I used Plasma SliderAnimation just because it was easier, but it could be replaced by a QPropertyAnimation
  • Since Plasma Animations expect to receive a QGraphicsWidget, I assumed that you are going to supply a QGraphicsWidget to the SliderLayout. But the real requirement could be a QGraphicsItem, since it is the position that is being animated.

You can get the code here, my personal sandbox for experimenting with plasma animations and other stuff (please, please, we *do* need to migrate away from svn and start using git) and look for src/lib/sliderlayout.cpp and src/example/slidertest.cpp. I wonder when the feature freeze of KDE is over, if this could be part of libplasma.

Maybe this concept could be expanded so there would be a layout class and the animation behavior (fade, rotate, slide, etc) be defined by a delegate. Anyone is tempted to help me to develop this further? ūüôā

July 4, 2010 at 9:05 am 1 comment

libgcal 0.9.5 plus a video

It is being a quite busy week and it shows since it is being just a few days since previous release of libgcal.

This is a highly recommended upgrade, and also I point every pacman (Package Manager/Maintainer) to use version 0.9.5 of libgcal to create packages for your distribution. As Fathi Boudra (the debian packager of libgcal) pointed to me, release 0.9.4 was missing 3 public user functions featured in previous releases.

Now ABI has being restored, plus some extras:

  • Support for IM field. This is not exported to akonadi resource yet, since it seems that in KAddressBook there is not an easy way to specify the IM type (i.e. skype, AIM, etc)
  • As said previously, restored ABI (gcal_contact_get_phone, _set_phone, _set_mail will map to the prefered field)
  • some memory leaks are fixed
  • improved unit test coverage

The akonadi resource has not changed, so you can continue using akonadi-googledata-1.1.0 *with* libgcal 0.9.5.

And last, but not least, special thanks to Holger Kral who is helping me to make KDE support for google contacts near to perfection.

To close this post, a nice video showing kaddressbook updating a google contact with basically all possible fields.

[blip.tv ?posts_id=3798837&dest=-1]

June 20, 2010 at 9:04 pm 7 comments

libgcal 0.9.4 released

Sweet! But what the heck is this ‘libgcal’ thingie? For starts, the name can and is misleading, since it should read as ‘library for google calendar’ but in reality, it implements both Contacts and Calendar google data protocol.

When I got it started, back in February of 2008, it was supposed to implement just calendar, but later on I realized that adding contact support only required +25% of code on it (thanks for well modular software design). Back then,  there was no other good alternative for any C/C++ programmer that would fit the following requirements:
  • easy to use;
  • well documented;
  • few dependencies;

So, I got my library started! After studying the google data protocol 1.0 (at the time) for while, I realized that using XPath would make my life way easier than say, browsing through the DOM tree searching for the attributes and tags that I wanted.

At that time, Qt didn’t have support for XPath (it only started with 4.5), so I went with libxml. For networking I used libcurl, which is fast/reliable and has great documentation (and a very welcoming community, from time to time I asked for help and always got answers for my questions).

From the very early beginning, I set a high quality standard in the development (after all, parsing XML in C is already prone to errors by itself) and followed a TDD (Test Driven Development) approach where *first* I write the test and *later* write the implementation of functions. Having an average of test coverage of 80% helped a lot when google released version 2.0 of the protocol (back in december 2008) and now more recently, version 3.0 of Contact’s protocol. I did the porting from version 1.0 to version 2.0 of the protocol in few hours, mostly because I could detect any regression by simply running the test suites.

IIRC in about 4 months I got the basic (authenticate, retrieve, add/edit/delete, query for updates) done and the library even got featured in the google official blog (that was surprising to say the least)!

So, the library was ‘done’, let’s put it to good use. I decided to integrate it with Opensync (it was rather cool, I got google contacts and calendar sync for my Nokia N95 over bluetooth working 6 months before google decided to release a syncml server for S60 devices). You can see a pre-jurassic video of this here. I think up until now libsyncml is a pretty good syncml implementation, it is just a shame that there is not a good UI bounded to it.

But I needed a good UI and the alternative seemed to write an akonadi resource. In just 3 weeks I got contacts working, while also implementing missing features in libgcal to make possible to do fast-sync (i.e. when you download only what has changed in server side). Contacts resource was done in just 3 days (I think this is clearly a good signal that akonadi API is well designed).

Developing the akonadi resources gave the opportunity to better understand how KDE community works and also to start running KDE trunk as my default desktop (after all, pre-packaged software are for sissies and developers should eat their own dog food).

So, why I wrote all this story? Well, to help to understand some numbers:
  • 10 months ……. since last release (0.9.3)
  • 6 distros ………. pre-packages libgcal (Debian, Ubuntu, OpenSuse,¬†Gentoo, Mandriva, FreeBSD) and counting
  • 2326 downloads .. directly from libgcal project website (hey, this is¬†a source code tarball of a library and not some porn!)
  • 20681 views …… reported by google analytics in a 1 year period
  • 3000 ms……… the lag to ping google servers in a bad day in¬†Manaus/Amazonas (yeah, truly it is unbeliavable how I managed to write¬†a *networking* library in this enviroment but what¬†doesn’t kill you makes you stronger)
  • 6762 LOC ……. lines of C code (34% are unit tests)
  • 76.5 % ……… current *real* code coverage (here I slipped a bit,¬†it used to be 80%)
  • 10th ……….. of most wanted KDE features

So, what this new release brings? For start, support for multiple email addresses, patch by Stefano Avallone (Andre Loureiro helped to fix the unit test) and migration to Google Contacts API 3.0. Next, support for structured names and several other fields (nickname, blog, etc) by Holger Kral. I think currently only IM field is missing from the library (but is quite easy to implement that).

You can have access to both the library and the akonadi resource in libgcal website. The only issue is that is required to purge your akonadi resource and do a slow-sync again because the ETags and urls of contacts have changed thanks to migration to version 3.0 of protocol.

So, what is missing in the library for a 1.0 release? The following features:
  • support multiple calendars (easy to do, is a matter of using another¬†URL as base to do the network operations)
  • support recurrent events in calendar (thanks to the fact that google¬†uses an invalid iCal to represent it, it gets tricky to implement it¬†since a iCal parser would fail to read this data). An idea to¬†implement this would be to ‘convert’ the invalid iCal from google to a¬†valid one and do the opposite when sending data back to the google server.
  • batch commit (nice to have, but not a hard requirement)
  • port/rewrite it all to Qt (seriously, this was actually started¬†already: http://code.google.com/p/libgdata-cpp/). Here I’m somewhat¬†unsure, if Qt has support for the XPath/XQuery in Symbian (it seems¬†that RTTI is not supported in this OS).

Oh well… so, why not give it a try? If you got the skills, go on and download the sources (please check the README and INSTALL files) and feel free to report to me how things worked (or not…).

If you are a normal user, I think in a couple of weeks it should get packaged for your loved distro.

June 11, 2010 at 2:16 am 19 comments

6 months of plasma in 10 minutes

Not really… It is actually 7 months of “kdelibs/plasma” (or libplasma) repository activity in 15 minutes (but that would not sound really as attractive as a post title).

Today a fellow co-worker showed to me a really nice application: gource (like ‘source’ but with ‘g’ letter, maybe for ‘graphics’) written by Andrew Caudwell. This application offers some really nifty source code control visualization and supports git/mercurial and other source code control systems using some hacks.

Since internet speed (not even talking about latency…) in Manaus/Amazonas is lacking to say the least, I decided to use git-svn to checkout a specific revision of kdelibs/plasma directory and later I did a rebase to update the code with the trunk.

So, for your amusement, enjoy.

[blip.tv ?posts_id=3669707&dest=-1]

I have some hints about this visualization:

  • in the first seconds there is an explosion of files (because I didn’t checkout the code from the very first commit in svn, is like the touched files warped inside of logs)
  • the code seems to be rather modular (there are a well defined set of clusters on it)
  • development speed is affected by some special dates (feature freeze, Christmas, Tokamak, etc)
  • there are a quite varied number of contributors working in libplasma (which is good! KDE is a pretty open community)
  • the overall champions in commits are (unsurprisingly!) Aaron and Marco ūüôā

May 21, 2010 at 3:42 am 2 comments

Lies, big fat lies and benchmarks

There is the old saying, ‘there are lies, big fat lies and there are benchmarks’, but after my previous post about the jsanim, it was time to try to measure how good or bad the javascript based animations are when comparing it against Plasma::Animations (written in C++).

== what to measure? ==

The first task was to have a meaningful number and frames per second (FPS) sounds like a good idea. The only issue is that QGraphicsView will only update parts in the scene that need redrawing (besides it seems that there is not an easy way to access the draw events without subclassing). My solution was to set a counter for the Animation::updateCurrentTime() method that is called while the animation is running, given that the documentation says that “… neither the interval between calls nor the number of calls to this function are defined; though, it will normally be 60 updates per second.”. Maybe I should call it Function calls Per Second (FCPS). ūüôā

Another important factor that can impact hugely is this kind of benchmark is both the hardware as also the graphics system used by Qt (and finally if the window manager has composite effect turned on). My test computer is a macbook (version 4.1), Intel Core 2 Duo (T8300@2.4Ghz), video Intel Mobile GM965/GL960), Qt 4.7 from master repository and I run the tests with composite disabled and got this results.

As the test application, I used the code available in playground/plasma/libs/jsanim, where a set of QGraphicsWidgets (each one drawing a pixmap) will run an animation group where 2 animations (zoom and fade) run in parallel followed next by a rotation.

== equal or not? ==

Next task is to extract any meaning from the results and I remembered what an old statistician professor that I had used to joke “Torture the data enough and it will confess”. I used to be good on this subject (back in the day I wrote some Bootstrap simulations and model fitting for biological data as also geostatistics¬† studies using R), but it is being 5 years since I last have done any statistical analysis, I was worried about making some mistakes while using more advanced methods.

So I went with the lazier approach and decided to use a boxplot, following bellow:

The js based animations seems to have a somewhat bigger variance (see second boxplot) but overall you can’t really tell if there is a difference between them (just in case, the p-values for t-student means tests was for each animation respectively: 0.54, 0.43, 0.86).

So… what about mobile? I did the same tests in a Nokia N900 (using the opengl and raster engines) and got this data, which yielded the following boxplot:

this time without the js variance and again without any sensible difference between js and C++ animations (p-values: 0.42, 0.16, 1).

== wrapping up ==

From the data (and after asking 4 different people to inspect visually how the animations were running in js X C++ apps), I can say that following the path of using javascript to code the logic of animations in Qt seems to be a valid approach with no sensible difference in performance. Kudos to QScriptEngine (and Qt development frameworks) for making it perform really fast.

There are indeed some issues in this study, following:

  • CPU and memory usage: I measure it informally using ‘top’ and they seems to be on the same level, but a more formal benchmarking ideally should measure this variables;
  • samples: I was a bit in a hurry and run each animation only 3 times (ideally a bigger sample like 20 should be used).

April 7, 2010 at 11:14 pm 5 comments

jsanim: current status

From my previous post about jsanim (javascript animations) where I explained the motivations about the idea, I’m writing to report the current status.

I got stuck while trying to create an instance of a javascript class from C++ side in QScriptEngine, but I was helped by khansen (you rock man!).

Being able to create an instance got to address an important issue in the first PoC: run more than 1 animation. Now each animation type will have a javascript class, where its instance will point to the animation target widget and hold the animation parameters. This design has some advantages:

  • uses a single QScriptEngine object (implemented as a singleton);
  • can run multiple animations;
  • multiple animations can access the same widget;
  • is more OOP oriented (the animation loop will call a method on the javascript class);

How fast (or slow…) is it to use QScriptEngine to animate a widget’s properties? What about a video?

[blip.tv ?posts_id=3433718&dest=-1]

Notes: it has 8 QGraphicsWidgets, 24 animation objects (zoom, fade, rotate), running in the Nokia N5800.

What is next is profiling and comparing it against our Plasma::Animation classes doing preferably the same type of work. But since it is still in the very early beginning, it is already looking promising.

March 30, 2010 at 2:00 pm 3 comments

Older Posts Newer Posts


Categories

  • Blogroll

  • Feeds