Planet PiTiVi

Last week was a blast. Spending a couple of days working on Pitivi fulltime and meeting with the awesome GStreamer folks again was a thrilling experience. Not only that, but it happened in the beautiful city of Málaga.

Coming from Montréal, I’m still a bit shocked at the sight of people wearing coats, scarves and tuques in broad daylight at +15°C.

Hell, the weather in Malaga was so consistently mild that even if we were in the middle of January, in the morning I went outside straight from the shower with wet hair, which dried in minutes!

While we’re talking about showers again… Malagans, unlike Bostonians, got the usability of their shower handles/lever/thingie right. However, they completely fail at road signage and urban planning. I thought I’d never see the day when I’d find something worse than the province of Québec, but it seems we have a new winner here:

• Street names written in minuscule font sizes on buildings (when they are present at all)? Check.
• Road signs inside the ramps/exits (when it’s too late)? Check. (Québec has that too)
• One-ways everywhere? Check.
• Inability to get back on a highway if you take the wrong exit? Check.
• Roundabouts where you have to yield to people outside the roundabout? Check.
• Roundabouts with streetlights every 30 degrees? Check.
• Tiny road signs inside the roundabout instead of on exits? Check.

I found this all quite amusing. Except when Antigoni and I had to get back to the airport and mistakenly ended up on an exit ramp on avenida de Andalucía.

Okay, back to the hackfest.

Notwithstanding the work we did on planes or busy airports and the many discussions we’ve had around tapas, we spent three days doing solid coding, debugging and ass-kickin’ on Pitivi and GES. I’m very happy that, in the process, Antigoni learned some new tricks and knowledge to make her more comfortable with hacking on the pitivi codebase. We also got our shares of laughs too. Like the fact that I ate three hours of Edward’s time trying to investigate why importing clips crunched my hard drive for many seconds… and then realizing that the culprit was not my code nor the gst discoverer, but GTK+ itself.

More precisely:

• Thibault spent the whole time hacking on GES and answering our questions.
• Antigoni went concrete/practical by attempting to fix undo/redo for effects, getting more familiar with GES in the process. I’m happy to have been able to answer some of her questions and being able to point out pitfalls in the code: at times, it even seemed like I knew what I was talking about, which is always a great thing!
• Edward spent nearly the whole time grunting and swearing in French, except when the sound was muffled by his palms:

As for me, I:

• Fixed image thumbnailing/permissions on the wiki and deleted 1846 spam image files scattered in 1539 folders.
• Reimplemented pitivi’s clip import process using the asynchroneous gst discoverer, which means that not only does the import progressbar work again, but it is blazing fast and doesn’t block the UI.
• Cleaned up code, standardized variables and fixed various bugs (such as seeking to the end of the timeline, or making the viewer check the pipeline position only when actually playing… your CPU will thank me)
• Implemented the ability to save/export the current frame as an image file. Hey, the code was just sitting there in GES, waiting to be used!
• Got convinced by Thibault to try implementing transitions and timeline video thumbnails myself. We’ll see how it goes.

We stayed one or two days after the end of the hackfest. Thus, on Saturday, pretty much the only day where weather sucked, the superhacker trio went on a touristic ride to Ronda, in awe at some of the alpine beauties of Andalucía:

More pics here.

This week made me realize/feel something even more strongly than before: since Thibault’s massive cleanup, hacking on Pitivi with GES is easy. No more files/modules confusion. No more “massive core” getting in your way. The code can still benefit from some cleanup/simplifications to make it feel more “pythonic” (patches welcome), but it already feels incredibly more agile and elegant. It now feels enjoyable rather than a maintenance ordeal. More like poetry, less like a thesis.

If you were hesitating to contribute to Pitivi, now is an exciting time to take a fresh look at it. We need help and there’s a lot of low hanging fruit that can be fixed. We’ll be happy to help you get started.

I’m currently in Málaga for the GStreamer hackfest. Hopefully, many bugs will perish. In the meantime, here’s a quick status update of stuff I’ve been doing in recent times in Pitivi.

• Cleanup the code for gtk actions so that the code is more readable and robust, fixing 629208 in the process.
• Cleanup the menus (again).
• Avoid having the viewer eating the CPU while idle.
• Fix various problems such as “Select unused clips” not working in treeview mode or reimplement removing all timeline instances of a clip when removing it from the media library.
• With the help of Brian Grohe, merge and delete spam user accounts on the wiki. I only have to fix image uploads and the https certificate issue now.
• Start fixing 629855; see below for details.

It turns out that pitivi was secretly writing autosave/backups of your project files (and deleted them when successfully saving/closing), but it did not actually use them. You’d have to know they exist, when they exist and where to find them, which is hardly intuitive.

So, I set out to fix this. Along the way, I ended up making the “unsaved changes” dialog more helpful. This is the result:

Then I worked on the actual dialog to prompt the user about the existence of an autosaved project file. It uses uses the same mind-blowing human-readable time formatting nekohayan technology, as you can see in the various following scenarios:

If you ignore the backup file, it will obviously be deleted when you cleanly save/close the project.

The big question is however, what we should do if the user selects to restore from backup:

• Should we simply rename the backup file to overwrite the “old” project file and then load it? This would mean the “old” project would be lost. In theory nothing should go wrong with that…
• If we want to be paranoid perhaps we could load the backup file as a temporary file and force the user to use “Save as” instead of a regular Save, but I’m not sure if that’s what users need. Thoughts welcome on this one, but I’d really like to have a simple solution.

Now that we’ve gotten rid of a ton of code in Pitivi (thanks to the port to GES), Thibault has been doing an incredible job at cleaning and reorganizing the remaining source files in a way that finally makes sense. I’m very happy about this: it means that it will not only make it easier for new contributors to get started, but also for regular contributors to not get lost in the various modules.

Behold, before (left) and after (right):

The old file tree was so big that it did not fit on a rotated 1920×1200 monitor, even with Nautilus fully zoomed out. Note that the tests/ directory is not shown here, it went from 33 test cases to 18 and now runs in one second (since the core tests are now in GES).

If you fail to be utterly amazed and emotional about what I’ve just shown here, I’ll make sure to explain properly the jawdropping significance of this cleanup (and the whole GES port in general), whenever I get a chance to make a talk at some hip conference.

Stuff I’ve been doing in general in the last three weeks:

• Fixed a couple of GES integration bugs in Pitivi. There’s still lots of work to do, but we’re getting closer to being able to make a first alpha release for your enjoyment.
• Redesigned the clip previewing feature. Instead of showing in the tiny timeline previewer, it now shows in a separate window that tries to show the clip at 1:1 size whenever possible. This window can be dismissed by clicking the close button or clicking anywhere outside the window, so there is no reduction of efficiency compared to the old approach.
• Merged, at long last, the power management branch. Pitivi will now inhibit the screensaver when playing and inhibit suspend when rendering.
• Make pitivi use Nautilus/Totem’s thumbnails. The result is prettier and we don’t have to do processing in most cases, which means even faster import/loading times.
• Pushed all that goodness to Pitivi’s origin (main repository) in a “ges” branch.
• Closed over a hundred old PackageKit bug reports on Freedesktop bugzilla. Hughsie should breathe a little bit more easily now.
• Finally killed all the spam pages (1000+) and images (500+) on the pitivi wiki, using the DeletePagePermanently Mediawiki extension (had to patch it for 1.17, see the talk page) and a horrible, horrible homemade python script to delete the pages automatically by opening hundreds of tabs in Firefox, accessing the “&action=delete_permanently” URL directly.
• The remaining problem is that I haven’t found a way to efficiently purge the remaining 800+ fake user accounts yet. Ideas welcome, the UserAdmin extension does not work with anything but Mediawiki 1.16 (and it seems unmaintained).

Me, fixing bugs. Or spending way too much time playing Vindictus.

In related news:

• At the end of January, I will be attending the GStreamer hackfest in Malaga, Spain. I predict that intense bug squashing will ensue. Come say hi if you’re around.
• Can someone please tell me if bug 666916 is a pygtk bug, a gtk bug or something I’ve done wrong? The same code was working a year ago. Current code here, offending line is #911.

I would first like to welcome Antigoni, who has been accepted as part of GWOP 2011 to work on PiTiVi. In the coming months, she will be helping Thibault fix bugs in our port to GES and, when we have a reasonably stable base to work with, implement the much awaited titling feature! She wrote an introductory post, but since it was published before she was added to Planet GNOME, it has completely flown under the radar.

I haven’t blogged much since the 0.15 release, partly because I was, you know, busy finishing* a graduate degree with 5 courses including 3 teamwork term projects and writing 13 case studies/papers. Nothing special.

Up until recently, I was thus unavailable for development work and Thibault could only work part time on the port to GES. Nonetheless, he has been steadily improving it. We now both have at least three weeks “full-time” ahead of us and a great intern to help us! As you can see, rumors of our death have been greatly exagerated. Our countless fans can rest assured.

Here’s one of the side-effects of development with Git: you don’t have to commit directly on master all the time, and so the lack of visible activity on “master” does not mean there is no development activity going on. Indeed, a lot of activity happens on github, and if you take Thibault’s “ges-port” branch into account you would see something like this:

Even when rotated 90°, my 1920x1200 screen does not have a high-enough resolution to fit the list of all the commits since 0.15! I’m not going to tell you yet how mindbogglingly huge and significant those changes are, because we’re not done yet and I want to keep some element of surprise (you may remember it is one of my chief weapons in my presentations). Let’s just say that this vastly simplifies development on pitivi. If we succeed, we will achieve world domination by the end of the world.

Now, you may ask, why hasn’t this been merged to master already? Because it’s quite unpolished and, in Thibault’s words, does not “deserve” to be there yet. I personally maintain a document listing all the bugs this branch has, and there are currently 15 of them. We may push to a branch in the official repository or merge directly to master as soon as we get the most severe bugs ironed out (hopefully very soon).

Your help is, as always, very appreciated. Feel free to chat with us on IRC about the state of the GES port and the way you can help us achieve our goal.

*: by the way, if you’re hiring and looking for a passionate open source contributor with a graduate degree in management (and a bachelor’s in psychology), feel free to poke nekohayo on IRC (or by mail, at gmail.com)!