Month: July 2007

Due to naively designing the internal organs of Tinymail’s async methods, I introduced a lot of threading and locking difficulties to the application developer who’s building an application on top of Tinymail.

We noticed this and therefore we did something about it (we is Rob Taylor and me). The solution was to introduce an internal queue that guarantees locking order for operations and during the connecting of accounts. The callback, which the application developer can implement, will make such a queue item waits for its completion. They happen in the mainloop, so that is done using a small little “GCond” construction. Dialogs like the password dialog, the SSL question and the code that gets launched in case the password was wrong will now all happen in the mainloop of the application. This makes it much more easy for the application developer to implement them using user interface components like the GNOME password manager’s dialogs.

This work was being done in this branch of Tinymail and will, with the migration of an E-mail client that is being build on top of Tinymail, be merged back to Tinymail’s trunk starting tomorrow. Some of the work has already been merged in trunk. Mostly work that was not related to the account connecting and initialization (which includes the password -and the SSL question things).

Hereby apologies for making these horrible internal design decisions. I guess I focused at the start of the project too much on the external design. I guess I hoped too much that internal organs would automatically become good too. This kludge made the external API behave difficult for application developers. It’s fixed now, and we wont make the same mistakes again. I believe this work is one of the key pieces in the puzzle that can turn Tinymail from a hobby project into an actual product.

The good thing about the kludge is that it forced me to make a lot of the locking both very fine grained and yet very correct. With the locking order in place now, this should mean that the overall quality should have improved a lot. We’ll see. Let the testers begin!

Another key piece in the puzzle is to someday have a libtinymail-dbus. We’re already planning and designing this, indeed.

Make gdk_threads_enter and _leave recursive locks. Is there a good reason why they are normal locks today?

This is the first release of libgee 0.1.0, the GObject collection
library.

libgee 0.1.0 is now available for download at:
   http://www.paldo.org/libgee/libgee-0.1.0.tar.bz2

libgee is a collection library providing GObject-based interfaces and
classes for commonly used data structures.

libgee provides the following interfaces:

        * Iterable
          * Collection
            * List
            * Set
        * Iterator
        * Map

The ArrayList, HashSet, and HashMap classes provide a reasonable sample
implementation of the List, Set, and Map interfaces. ReadOnlyCollection,
ReadOnlyList, ReadOnlySet, and ReadOnlyMap are read-only wrapper classes
that prevent modification of the underlying collection.

libgee is written in Vala and can be used like any GObject-based C library.
It's planned to provide bindings for further languages.

Development is going on at svn://www.paldo.org/libgee/trunk

        Jürg Billeter

Dave writes on his blog an interesting point of view on why mobile device vendors should stick to the open standards. He’s on a blogging spree, so make sure to read the follow-ups on this item too:

This also highlights a serious issue with closed and non-standard platforms. If this were an ordinary smartphone, you could just shrug, ditch the in-built email client, and use something different. I’ve no idea whether, for instance, the email clients on the 770 and N800 are safe to use or not – but if I were using them, and I found it was insecure, I could switch to a different one. Security issue solved. I can switch to a different client because anyone can – in principle – write one, because the platform is open (allowing them to write programs for the platform) and the standards are open (allowing them to write an email client).

With Apple’s choice to use a proprietary system with Yahoo, the option to be secure has been denied to their users.

Had Apple and Yahoo chosen to use the existing, open-standard, Lemonade protocol suite, this simply couldn’t have happened.

ps. Tinymail and Modest of course use the standards, securely.

Although I’m one of those guys who’d like to see all platform code to be run by a jitter that can support all the programming languages, platforms and devices that we use today and in future, I got a very nice second introduction to Vala at GUADEC.

The nice thing about Vala is that it gives you a programming language that looks and feels like C# and D (generics, interfaces, properties, classes, delegates and all that in a syntax that looks almost exactly like C#), yet it’s completely compatible with GObject and GTypeInterface. That’s because the Vala compiler behind your back first compiles to a GTypeInterface or GObject before passing it to your normal C compiler.

Which sure sounds scary. Though … if you think about it for longer than a few seconds, it’s absolutely not a very bad idea at all. Especially not when it comes to integrating with existing GObject infrastructure. When it comes to having a nice higher programming language where your software’s release tarballs need nothing more than a C compiler and glib with gobject.

Vala is smart enough to also do this right, to generate clean GObject code and to allow inheritance and implementing from existing GObjects and GTypeInterfaces. It manages to get generics and reference counting to be just right (reference counting is done using smartpointer likes. It’s done for you automatically, so there’s no funky C++ syntax involved. You also have weak references to avoid embraced references).

The author is also working on automatic language binding generation to programming languages like Python and .NET. This is more easy with a Vala programmed piece of library because the Vala compiler knows from the Vala syntax that the developer did everything in a certain way (and not in one of the ten million ways that the C programming language allowed him to work).

On top of that is Rob Taylor working on his stuff to parse the C tree of a piece of C header code to an XML file. For this XML is useful because on that resulting XML we can apply XSL stylesheets. For example one that formats it into something that we can embed into the shared object file of a GObject library, to add introspection data. Or another XSL to aid a tool like gtk-doc to create an indexable API documentation database, HTML or whatever file that can then be used to be put online or to be used in tools like Devhelp or for code completion in an IDE like Anjuta.

But also something that can be utilized to fully automatically generate completely correct language bindings. On dynamic or so called “always late binding” programming languages, like Python and PHP, it could even do this at runtime. With type safe programming languages it could at compile time use this new and extra information to without any real help from a human developer, create a completely correct language binding.

Since today for both POP and IMAP it’s possible to, while you are retrieving summary items, start opening (downloading) uncached E-mails. On IMAP this is done (a few weeks ago) using an extra socket. With POP this is done (today) by throwing RETR commands in between the TOPs that are needed for the summary fetch on the same connection (most POP servers lock the access to one connection per user). It’s also possible and well tested to have Push E-mail events of 1000ths of new and expunged E-mails. It’s possible to see your headers (summary) view to get filled up while retrieving summary info. All this while Tinymail attempts to keep the bandwidth acceptable low for GPRS.

Jose Dapena Paz also recently added support for removing attachments from already cached E-mails. This is a feature that is interesting for devices that don’t have a lot of (flash) disk space. Again retrieving the uncached attachment should also work. Partial retrieval (initially retrieving without attachments) is working, but suboptimal. We’ll be working on improving features like this soon. POP is of course much more difficult as there’s no way to know where a MIME part starts nor to select which of the lines you want to retrieve.

Huge amounts of small and big things are being fixed. Required features being implemented. A lot of work is being done.

By changing the way work is done, electronic networks may lead to a new kind of economy centered on the individual. — Thomas W. Malone and Robert J. Laubacher.

In business, I believe the future lies in loosely connected individual contractors, dynamically coming together to attack a project or idea. — Rob Taylor.

When Rob blogged about it, I decided to order my copy of that document. I just read it. Sounds good to me.

Let’s talk about it at GUADEC. I will arrive at BHX the 15th around 13:50 by the way.

I would, out of the blue, like to highlight the work the people doing the Anjuta 2.x series have done. I have been working with whatever is in Subversion (and before that, CVS) for multiple years now. It had its weeks of total instability, weeks of total stability too. But on features, especially on features, it has seen vast improvements. It’s a totally different experience when compared to the 1.x series of Anjuta.

pvanhoof@schtrumpf:~/repos/gnome/anjuta$ sudo make install
...
 /usr/bin/install -c -m 644 'anjuta.desktop' '/opt/anjuta//share/applications/anjuta.desktop'
make[2]: Leaving directory `/home/pvanhoof/repos/gnome/anjuta'
make[1]: Leaving directory `/home/pvanhoof/repos/gnome/anjuta'
pvanhoof@schtrumpf:~/repos/gnome/anjuta$ 

They are now working on integrating some bits and pieces with scratchbox. The integration with autotools (which utilizes gnome-build) is unimaginable good. I’ve become a total addict of features like the symbol browser and the symbol searcher. I’m sure I’m missing quite a lot of other features that could potentially glue and duck tape me forever to Anjuta’s 2.x. A requirement for any distribution that I’ll use has been, since more than a year, that I can easily build whatever I find in Anjuta Subversion -> “apt-get install gnome-devel”. Thank you, Ubuntu and Debian, for that.

I believe we need more of our developers to use this project. That way we can get this project to become a product, too. Because frustrated developers will fix Anjuta’s code. In contrast, I haven’t seen a lot of GNOME-developer’s frustration get converted into code for the Eclipse CDT project. I’m confident that since Anjuta is purely GNOME/Gtk+ style of development, Anjuta would be different.

I often got so pissed at whoever broke the search dialog of Anjuta’s Subversion this time. Forcing me to go back to ‘grep’ and ‘find’, to find things. Slowing me down. But then again, I should just help them in stead. Right? In stead of remaining pissed, I’ll truly apologize for being a user of their software yet not having contributed a lot to their fabulous 2.x series. On the bright side, I did do some work on the file dialog of the 1.x series of Anjuta. That code has now vanished and been replaced by superior things, of course.

I’ll put an icon on my desktop that throws Anjuta’s process into gdb. Being a developer you’ll be surprised how often you click it and then actually fix that annoying bug in stead of just restarting the software. I used to do this with Evolution too.

We’re getting there. Although I’m thinking in months, not in weeks. If you’re working on a feature for Tinymail and you want it in its first release: you better start hurrying up.

It’s becoming a product that just works. A lot of the release work will be low hanging fruit like getting gtk-doc in perfect shape, making sure ‘make distcheck’ does the right thing, killing a few more major defects and writing some more documentation.

When those items are finished, the project can be delivered as a product. Above all, it must earn the status “product”. That’s the point where I’ll release a first version.

Rob (Taylor) used to be one of the lead developers of the Telepathy project, he co-founded Collabora and is now starting up Codethink Ltd. He has worked on HAL, D-BUS, Telepathy and various other bits and bytes of the many GMAE and GNOME components.

Although not the only participant in this, I believe he played an important role at encouraging a number of companies in working with us, the GNOME Free Software community. He combined his technical expertise with business expertise.

It’s therefore my opinion that Rob has become one of our most skilled and needed people. At LinuxTag we agreed to start working together under Codethink’s flag. I will be the first contractor to come on under the new Codethink plan.

With our expertise we are planning to change the mobile landscape. We’re both passionate people who’s goal is to make the difference, which is exactly what we’ll do.