Tag Archives: Openismus

gtkmm 3.0.0

General, ,

We’ve been working on gtkmm 3 increasingly since sometime in 2009 and today we have finally declared it stable. gtkmm 3.0.0 is out. We last did an ABI-breaking parallel-installable stable version in April 2004 for gtkmm 2.4.0 so we’ll probably have to live with this for a while.

We didn’t fullfil all our amibions, due to lack of time, but we did fix many annoyances that had built up over the years. And luckily GTK+ 3 turned out to contain far more interesting changes than initially planned, most of which are now available via gtkmm-3.0.

Personally, I’m glad that it’s done. It’s been a long hard slog over the past year, made possible by my company Openismus GmbH, just to keep up with the (nevertheless necessary) API churn during the unstable GTK+ 3 development cycle. That hasn’t left me much time for any other software development, but now I feel free.

To port to gtkmm-3.0 you should first try to build your application with gtkmm-2.4 but without using deprecated API. The deprecated API generally has documentation suggesting what new API you
should use instead. These gtkmm-3.0 porting hints might also be helpful:

Some new/changed API in gtkmm-3.0, though this list is far from complete:

Gtk:

  • Adjustment is now used via RefPtr.
  • AppChooser, AppChooserButton, AppChooserDialog.
  • Box, ButtonBox, IconView, Paned, ProgressBar, ScaleButton, ScrollBar and Separator now derive from Orientable.
  • All widgets derive from Buildable.
  • Several widgets derive from Scrollable.
  • CellArea and CellAreaBox
  • ComboBox now derives from CellLayout.
  • IconSet is now used via RefPtr.
  • StyleContext, StyleProvider, and CssProvider, replacing Style and Rc.
  • Grid
  • NumerableIcon.
  • Switch
  • Widget::on_expose_event() is now Widget::on_draw().

Gdk:

  • Added DeviceManager.
  • Removed Drawable, Pixmap and Bitmap.
  • Cursor is now used via RefPtr.
  • RGBA replaces Color, though Color still exists because it is used by TextView. We hope to deprecated Color completely in gtkmm 3.2.
  • Removed Colormap and its general awkardness.

General:

  • Use std::vector in several methods instead of the intermediate *Handle types to make the API clearer.
  • Change all operator void* to operator const void*, with a BoolExpr typede and extra documentation. This avoids an unintentional implicit cast to int, and is generally considered good practice in C++. The new C++ standard allows the use of the explicit keyword here instead, but that is not yet widespread.
  • Many existing classes have many new methods.
  • gtkmm-3.0 currently has no deprecated API. Not even API that was added and then deprecated during GTK+ 3.0.

See also the list of new API in gtkmm 3.0.

Thanks to everyone who helped.

glibmm 2.28 with Gio::Settings and Gio::DBus

General, ,

We finally released glibmm 2.28.0. This is the first stable release with new API since glibmm 2.24.0 in March 2010. We skipped glibmm 2.26.0 because there was so much new and awkward API that needed to be wrapped properly for C++. It’s scary to declare so much new API stable, but we had to do it eventually.

Glib has the following new API, though this list is not complete:

  • Variant: A new hierarchy of templated Variant<> types, for use with Gio::Settings and Gio::DBus.
  • Regex uses a new MatchInfo class.
  • build_filename() now has overloads to take up to 9 arguments,
  • get_system_data_dirs() and get_system_config_dirs().

Gio has the following new API, though this list is not complete:

  • Settings: For application settings, replacing GConf (or Gnome::Conf).
  • DBus: API to use or implement D-Bus services.
  • Proxy, ProxyAddress and ProxyResolver.
  • SocketControlMessage, UnixCredentialsMessage and UnixFDMessage.

See also this list of new API in glibmm 2.28.

Gio::DBus, Glib::Variant and Gio::Settings were a huge amount of difficult work mostly by José Alburquerque, with help from Jonathon Jongsmam, Yannick Guesnet, Michael Edwards, Martin Braure de Calignon, Murray Cumming and others. Some of José Alburquerque’s work, and most of Murray Cumming’s was funded by Openismus GmbH.
Thank you, everyone.

Glom: GtkNotebook with no frame

General, ,

GtkNotebook has a frame around the page that it shows, grouping that page’s widgets together. That makes sense when the notebook is part of a more complicated UI. But Glom uses a GtkNotebook for almost all of its main window, taking up all of the window’s width. The frame needs an inner border that wastes space and the nesting actually makes the UI look more complicated than it is.

That frame is not optional in GtkNotebook, so I had to create a custom Gtk::Notebook-like class to replace it in my code. The result is mostly better, though I don’t like those two toggle buttons much. Suggestions are welcome.

I will also move those top labels around, putting them in the same row as the List and Details toggle buttons, but this is a start.

Return of the GNOME Events Box

General,

Retrieving the Box

In November I mentioned that UPS had lost the GNOME Events Box. Well, we have retrieved it after we did the investigation that UPS would not do, or did not want to tell us about. I didn’t have time to chase this myself, but I want to thank these people:

Henrique Rodrigues personally went to Odivelas in Portugal and tracked down the box’s physical location. He learned that the (incorrect) receipient had at first assumed that it was one of their regular deliveries and had broken the locks to check what was inside. When they discovered that it wasn’t for them, they told UPS, requesting that they take it away again and give them an official acknowledgment that they had not stolen any of the contents. UPS said No and then told us that they couldn’t retrieve the box and that we should send a claims form with all our receipts. The Box remained in a warehouse belonging to the incorrect recipient.

With this information, Bastien Nocera and Ekaterina Gerasimova contacted UPS repeatedly and in various ways, until they agreed to pick up the box and bring it back to Berlin, where it has now arrived.

UPS are usually very efficient and better than the alternatives. But this was all rather silly.

Looking for a new GNOME Events Box Host

I don’t have the time these days to properly manage the GNOME Events Box, so I’d really like someone else to give it a home. You just need to store it, arrange for UPS delivery, and receive it when it returns. I have generally paid peoples’ expenses when they send it back, and I then ask the GNOME Board for reimbursement for all receipts every now and then.

In the meantime, Canonical have generously paid for (and Bastien organized) a replacement GNOME Events Box, so we now have two. So you might end up hosting both, though I guess that the GNOME Board might choose to send one to another part of the world.

Avoiding Use of Deprecated API

General, ,

Most Deprecated GTK+ 2 API is not in GTK+ 3

GTK+ 3 has API changes compared to GTK+ 2, but some of the new API has been added to GTK+ 2.22 and 2.24, and older API has been deprecated. gtkmm has done this too. This is intended to ease porting to GTK+ 3 (or gtkmm 3). If your application builds with GTK+ 2.24 with deprecated API disabled then it will be much easier to port to GTK+ 3 afterwards.

This requires defining C macros such as GTK_DISABLE_DEPRECATED, though you’ll want to use GDK_DISABLE_DEPRECATED, GDK_PIXBUF_DISABLE_DEPRECATED, G_DISABLE_DEPRECATED and maybe PANGO_DISABLE_DEPRECATED too. You can specify these as -D options in your Makefile.am.

However, you should not disable deprecated API in your build at all times. That just makes life unnecessarily difficult for people building from tarballs.

So over the last few years, Daniel Elstner has perfected an MM_ARG_ENABLE_WARNINGS m4 macro for use with autotools. It adds an –enable-warnings=min/max/fatal/no configure option. We use it mostly to turn on compiler-warnings-as-errors, but I think that it would be very useful to many people now during the transition to GTK+ 3.

MM_ARG_ENABLE_WARNINGS

This macro is in mm-common, which you could depend on, though it contains other stuff that is only interesting to gtkmm projects. For instance, I use it in Glom’s configure.ac file, like so:

MM_ARG_ENABLE_WARNINGS([MYPROJECT_WFLAGS],
  [-Wall],
  [-Wall -Wextra -Wno-missing-field-initializers -DGSEAL_ENABLE],
  [G GDK GDK_PIXBUF PANGO GTK])

I then use the resulting variable in my Makefile.am file, like so:

AM_CFLAGS = $(MYPROJECT_WFLAGS)

If you use non-recursive autotools (and you should) then you won’t need to repeat that much.
I also added the option to DISTCHECK_CONFIGURE_FLAGS so I am forced to fix any warnings during make distcheck, when doing a tarball release. You don’t need to do that.

Whenever I build one of these projects from git I specify –enable-warnings=fatal to autogen.sh and fix any problems that it finds. That’s why there are zero compiler warnings or use of deprecated API in Glom.

DK_ARG_ENABLE_WARNINGS

You might prefer to copy the standalone version into your project instead. There are copies of the standalone macro in various projects already. Here is an example commit that adds it to a project.

The macro call looks like this in your configure.ac, for instance:

DK_ARG_ENABLE_WARNINGS([MYPROJECT_WFLAGS],
  [-Wall -w1],
  [-pedantic -Wall -Wextra -w1],
  [G GDK GDK_PIXBUF PANGO GTK])

Translated gtkmm.org Web Site

General, ,

Since May 2010, gtkmm.org has been available in Chinese as well as English, thanks to a patch from Tao Wang. It uses DocBook XML and xml2po, like any other GNOME documentation, along with some custom DocBook XSL stylesheet parameters and our web site’s CSS.

It works. Translations can be kept up to date easily when the original English changes (though they haven’t so far in this case, admittedly) because the translations are not in copies of the documents, but in regular gettext .po files. It’s not a CMS, and editing is not particularly user-friendly, but it’s no worse than the untranslated HTML that we had previously.

I recently cleaned up the web site’s directory structure so it’s more like an application or documentation manual so you can see what I mean. And the gtkmm.org translation status is available on l10n.gnome.org thanks to Claude Paroz. The gtkmm book is also translated in the same way, so now everything other than our (Doxygen) API reference is translatable.

The build could be improved by someone who knew how. You currently need to add an HTML_generation block in the Makefile.am when you add a new locale.

Trainee mini gtkmm projects

General, , ,

In August I mentioned the mini GTK+ C projects that our trainees completed: massifg, gmemory and GHangTux.

They recently finished some mini gtkmm C++ projects too, because I think gtkmm is a comfortable way to learn good C++ habits after GTK+. Here they are:

  • Chris Kühl’s Merkmal, a port of gmemory to gtkmm and cluttermm. It’s a card memory game.
  • Patricia Santana Cruz’s GHangtuxmm: A port of GHangTux to gtkmm. It’s a hangman game.
  • Jon Nordby’s glom-postgresql-setup, which does the awkward server configuration that typical Glom users don’t want to worry about.

They are now moving on to Qt and they’ll get some exercise on deeper OO and C++ concepts. In fact, Jon Nordy has already created a QImage plugin to support the OpenRaster image format.

Openismus contributions to GTK+

General, , ,

Openismus has made several contributions to GTK+ recently, getting changes into the upstream releases, doing things properly via bugzilla and the mailing lists, working with other developers even when it’s complicated, so it’s done right. This has helped my application, Glom, which uses GTK+ via gtkmm. It shows that we could do the same for you.

For some of our employees this is also a slight reward for being so busy coding with Qt right now. We love GTK+ as much as ever, regardless of one big company’s arbitrary choice to stop using it. Some of us like it even more now.

A short list of recent Openismus work on GTK+:

Tristan also created a couple of useful container widgets that were not needed in GTK+ itself so they are now in libegg. They were made possible by the extended layout support in GTK+ 3.

  • EggWrapBox
    By Tristan. This positions child widgets in sequence according to its orientation. For instance, with the horizontal orientation, the widgets will be arranged from left to right, starting a new row under the previous row when necessary, as in a toolbar or tool palette.
  • EggSpreadTable (screenshots)
    By Tristan and ported back to libegg by our David King. This positions its children by distributing them as evenly as possible across a fixed number of rows or columns, like newspaper columns.

gtkmm 2.91.2

General, , ,

Just a day after the GTK+ 2.91.2 release, I managed to release gtkmm 2.91.2, with (hopefully) all the new API. I have Glom and the gtkmm-documentation examples building with the latest gtkmm as a test.

I also did a release of glibmm 2.27.1, matching the glib 2.27.1 release. José Alburquerque has been working very hard on this, while also working on gstreamermm. But we really need some feedback on all the new API, particularly the Gio::DBus* stuff. Someone needs to sit down and write example code to see if it even works. We skipped the stable glibmm 2.26.x releases because this had not happened so we were not ready to call the API/ABI stable.

libgda’s SqlBuilder API

General, , ,

SQL is complicated and strange and subtly different depending on the server, particularly for complex queries, which means anything that a user might find useful.

Glom built SQL queries internally, concatenating strings together with some helper functions. But that was still fragile and repetitive so I recently ported Glom to use the new GdaSqlBuilder API, via its Gda::SqlBuilder C++ wrapper. I don’t know of any similar open-source API for this in C or C++.

GdaSqlBuilder doesn’t completely hide the almost-free-form structure of SQL, but it does encourage you to only write queries that makes sense. As a bonus, you don’t have to worry about correct quoting, correctly representing values as text, correctly escaping that text, or correctly quoting table and field names. It should also hide most server-specific syntax, making it easier to port applications to different database servers. I’m happy to remove some of that awkward code from Glom.

Vivien Malerba was very responsive to my feedback, so I now think that the GdaSqlBuilder API is about as good as it can be in C. I think Gda::SqlBuilder is even nicer, thanks to C++ method overloading.

But it still doesn’t feel quite right – you can’t guess what the functions do without reading the documentation, many functions have similar names, and some functions can only be used on certain types of queries at certain times. I guess the Select and non-Select queries should be in separate (related) classes, instead of just mentioning “select” in the method name, but that would lead to annoying casting in C. Maybe I’ll do that in the C++ API.