It’s been a good first day at the first ever West Coast Summit in my native city of San Francisco. A good turn-out all around and a lot of interesting talk about what’s next for GTK, particularly under-the-hood.
When Yorba announced Geary some time back, one question we were often asked was, Where’s the calendar? When we replied that Geary was not going to have a calendar — that it was an email client, nothing more — the question often turned into a request: Will you make a calendar app then?
We saw it coming. When we were planning Geary, we talked internally about calendaring and its close relationship to email. We agreed that, like our approach to Geary, that was best handled by a dedicated application rather than bolted-on as yet another feature.
So it’s not terribly surprising for me to announce that Yorba has begun work on a new calendar application for GNOME 3: California. The philosophy that guided our development of Geary is present in California as well: simple to set up, quick and easy to use, focused on its particular task but flexible for each person’s workflow. The first commit was in January and work has progressed far enough that we’re ready to start talking about it.
The current implementation relies on Evolution Data Server (EDS) for all backend calendar operations, but the code is architected so other backends (such as GData) can be added later, perhaps even broken out as plugins. The design is to use most of the modern GTK widgets and design goals, including GtkHeaderBar. A lot of features are still unimplemented, of course, but development is continuing.
For more information and links, visit California’s home page. Our ticket list is a good way to see what’s on the immediate horizon and report bugs and feature requests. You can also subscribe to the California mailing list to discuss the application’s current state and future.
Be aware that in its current implementation you’ll need another EDS client (i.e. Evolution) to add and configure the calendars visible to California, but we hope to soon have that implemented as well. And, no, Geary and California don’t talk yet, but that’s in the works.
What about GNOME Calendar?
One question we’ve already been asked is, why not work on GNOME Calendar? We have a few reasons.
First, GNOME Calendar is written in C. I know it sounds like language-chauvinism, but that’s not the language we at Yorba feel should be the future of GNOME application development. (That is, the language for developing full-blown applications; I’m fine with C for the underlying libraries, although I would still suggest library developers consider Vala.) It’s not fear of C itself, it’s the boilerplate and bookkeeping that GObject C development requires that becomes more and more onerous as a project grows. It’s also the ramp-up required for new contributors to make productive patches. We have coded two mature applications, Shotwell and Geary, in Vala. We have good reasons for coding a third in the same.
Second, I have particular opinions about the underlying design (not merely the user interface) for a network calendar application that I felt were not being approached in GNOME Calendar, or indeed in other calendar applications we looked at. With California I’ve put a bit of work into a flexible date and time model that treats date spans as iterable collections; makes various units of time (weeks, months, years, etc.) iterable date spans as well as concrete units; knows the difference between calendar time and wall clock time; can deal with iCal‘s timezone model; and so forth. The date/time model is cleanly separated from the backend network connectors and from the UI itself. It also makes full use of GObject’s properties and signals so they can be directly bound to GTK widgets. This is a GObject application; let’s live and breathe GObject then.
Third, the knee-jerk reaction that everyone in the GNOME community should be working on a single set of applications is not realistic nor a vision for growth. There’s nothing wrong with GNOME users having choices for their applications. Parallel development of applications even fosters growth as one team takes ideas from another, and vice-versa. Some people like Thunderbird, some people like Evolution, some people like Mutt, some people like Geary; what’s wrong with that? I suspect the idea is that it’s better to pool all available resources behind one code base, but that assumes a flawed one-size-fits-all vision of software that is less true in 2014 than it ever was before.
Yorba has released Shotwell 0.18.0, the latest version of our GNOME photo manager. Updates include:
- Tumblr publishing fixed
- Fixed resource leak when importing videos
- Improved searching for event comments and diacritics
- Various bug fixes
- Compile and install translated help files
- Updated translations
We at Yorba are pleased to announce the release of Geary 0.6.0, a new stable version of our IMAP mail client. Much has changed since our last stable release, including:
- Basic search scope operators, such as from:john
- An oft-requested “save sent mail” option for non-Gmail users
- Separate actions for archive, trash, and delete
- Much improved server support — if you found Geary unusable in the past on a Dovecot or other non-“cloud service” server, I’d urge you to give it another go
- Many stability, speed, and interface improvements and bugfixes
Last Friday Daniel Foré made an announcement on elementary’s mailing list that has generated a considerable amount of interest:
Shotwell needs a new maintainer. … After some discussion with Jim, we think the best course of action is for elementary to fork Shotwell.
OMG! Ubuntu reported later that day (and with surprising speed) that “Elementary To Take Over Shotwell Development”. I’ve received emails from people wondering what this means for the future of Shotwell.
To be clear, Yorba remains the maintainer of Shotwell. elementary has not taken it over. Yorba recently took the rather large step of moving Shotwell and all its associated work (including our ticket database, wiki pages, architecture guidelines, and more) into the GNOME infrastructure. We didn’t do all that work just to hand the keys over to another group on another platform. We made the move because, when we did all the calculus, we agreed GNOME was the best location for the future of Shotwell. Going into the GNOME infrastructure opened (and is still opening) a lot of doors for us. However, Yorba’s resources are limited, and as of late we’ve not been focused full-time on improving Shotwell.
This put us into a bind, and so I began looking for outside individuals or groups who would be capable and willing of contributing to Shotwell in a more substantial way. I kept coming back to a group that has shown a great deal of energy and motivation in the past and a willingness to work with us: elementary. They expressed a lot of interest in improving Shotwell, and so we began a discussion about how that situation might look. In the end, we reached an understanding:
- Shotwell will continue to be maintained by Yorba. We’ll triage bugs, take patches, fix critical bugs, ensure it builds with the latest versions of Vala and supporting libraries, and runs on the major distros. Yorba will continue to release Shotwell on a six-month schedule.
- elementary agreed to jump in and improve Shotwell, but stated that all of their development must occur on Launchpad. They also wanted to rebrand the project to fit under their design umbrella. This includes integration with Contractor and other technologies they’re building for their OS.
- It was agreed that they would fork the project to Launchpad and rename it Pantheon Photos to distinguish it from Shotwell. The pragmatic reason for this is to prevent name collisions with packaging. This also allows for elementary to customize Shotwell to their own platform and brand it separately.
- As Shotwell maintainers, we’ll evaluate elementary’s work and look for commits that are useful to Shotwell users. We’ll cherry-pick that work and merge it into the Shotwell base.
- elementary agreed not to relicense Shotwell, so its current license (LGPL 2.1) stands.
While elementary has big plans for Pantheon Photos, none of that work is happening in private and we’re free to take improvements as we wish (and likewise they’re free to pull improvements from Shotwell’s code base).
Additionally, this situation already exists, and has existed for years, with Ubuntu: they maintain their own fork of Shotwell to provide integration with Ubuntu Online Accounts. (I’ve referenced that situation before.) elementary is doing this on a larger scale, admittedly, but it’s not unique in the world of open-source and shouldn’t be a source of alarm.
(I regret that Daniel wrote “Shotwell needs a new maintainer.” I was privy to that email before it was posted and should’ve asked he reword it. That was my fault, not his.)
In October I briefly announced that Yorba was virtually moving — that we were in the process of transitioning from our self-hosted project server to GNOME’s infrastructure. Well, the move is all but complete. I’m pleased to announce that we’re now officially hosted inside the GNOME ecosphere: Bugzilla, git.gnome.org, mailing lists, the works. Here are links to our new wiki.gnome.org pages:
First, let me clarify that the Yorba Foundation has not merged or become subsumed by the GNOME Foundation. Financially and legally we’re still independent entities, albeit with a shared goal of spreading free software far and wide.
So, why’d we do it?
Yorba has always been a GNOME developer. Our applications are GNOME-specific and we build them with GNOME technologies and libraries. For almost five years now we’ve enjoyed a certain independence developing from outside the GNOME infrastructure, but we also were giving up a certain amount of integration that perhaps would’ve made our offerings that much better. It’s a trade-off. Recently we re-evaluated those trade-offs and decided that it made sense for Yorba to move into a tighter GNOME orbit. As such, we look forward to working closer with the GNOME team on things like application design and the future direction of the free desktop.
Yorban Charles Lindsay did a great job moving a mountain of our data to gnome.org, in particular our voluminous ticket database, which is approaching five years old. Not only are all our tickets now imported into GNOME’s Bugzilla (even their attachments), the final entry in our legacy Redmine tickets include instructions on how to locate that ticket in GNOME Bugzilla. He automated much of this with a Python script we’ve made available on github in case someone else out there needs to move their tickets (“issues” in Redmine parlance) from Redmine to Bugzilla.
I’d like to extend deep thanks to GNOME sysadmin whiz Andrea Veri for making the transition about as painless as I could imagine — no, even less. Andrea guided us through this process with grace and steady vision. I’d like to echo William Jon McCann’s call to “ensure Andrea Veri’s contract is renewed and stays on as the GNOME sysadmin”. You have Yorba’s vote, av!
New version number scheme
In the past, Yorba used a standard sequential versioning scheme (0.1.0, 0.1.1, etc.) To differentiate between stable releases and builds from source control, we appended a descriptor to the version number (0.1.1+trunk, 0.1.1+branch). When we needed to issue pre-release tarballs to our partners, we used a different descriptor (0.1.1pr1, 0.1.1pr2).
Over the years we received a number of complaints about this, especially from package and repository maintainers (whose business, if you think about it, is very much about version numbers). With our move to GNOME, it seemed like a good time to revisit our scheme. The one numbering system we kept hearing was worth adopting was GNOME’s itself, where even numbers represent stable releases and odd numbers represent unstable releases. (I believe GNOME adopted this from the Linux kernel; where they got it from, or if they invented it themselves, I don’t know.) In addition, it was requested that we release unstable tarballs more often, to give distributors and packagers more time to prepare.
We decided to make the switch and have already adopted this with Geary. We recently released an 0.5.0 unstable, and will follow up soon (probably after the holidays) with an 0.5.1, and so forth. The next stable release of Geary will be 0.6. Similar changes will be coming soon for Yorba’s other projects.
Translations / internationalization / localization
One goal we have is to move all our projects that require internationalization into GNOME’s system. We’re not quite there yet; we need to do a little more work on our build systems in order to integrate fully. I hope to get that wrapped up next month. If you’re eager to offer translations for our projects, all I can ask is for your patience until we get this resolved. (If you’re a GNOME internationalization whiz, we’d appreciate your help in finishing the work.)
Due to this expected move, I’ve closed our Transifex project repositories. If you’ve made some translations recently, don’t worry, they’re not lost, I simply haven’t merged in the final snapshot of those strings. I’ll do so after the holidays.
Even though we no longer use it, we’ve decided to keep our Redmine project management server up for historical reasons. We’ve locked it down to prevent any changes.
With the server transition we also decided to clean house a bit. Over the years Yorba has developed or adopted technologies that didn’t get off the ground. Although Yorba is no longer maintaining them, we didn’t want them to simply disappear, as so we’ve moved them to our public github repo. The projects I’m speaking of are media (our early attempt at audio/video editors) and Valadate (a Vala-based unit testing framework).