Monthly Archives: March 2009

EclipseCon notes and git

Well, I’m back at my day job at Wind and am doing a little reflection on what happened last week at EclipseCon. Despite the rumored lower attendence, I met with pretty much as many people as I do every year, maybe even slightly more. And hopefully that’ll translate into growth in the CDT community.

And I guess my talk on building communities was a little over the top on the subject of project “takers” that I had a number of people come up to me and apologize and offer to contribute in the future. I certainly didn’t mean to offend or criticize. I just wanted to prepare project contributors that there are vendors and people who are happy to take your work for free and not give anything back. And, while that’s frustrating, with open source licensing there’s nothing you can really do about it but be mentally prepared to see it happen.

The one area where I saw a very interesting rise in momentum was with distributed version control at Eclipse and git in particular. It started with discussions at the councils on Sunday where it became obvious that we need significant community interest and willingness of the projects to drive this home to justify the significant cost to the Foundation. There were a couple of BOFs and panels that I missed but it sounds from Karl’s post that the activity was significant enough for the Foundation to start putting a plan together. Very cool!

The more I talk to people and dig into it, I think git and DVCS in general is probably one of the most significant technologies driving change in the way we work on software development projects since objects invaded our world. As evidence of that, the thought of distributed source repositories opens the door to the IDE in the Cloud. The Cloud could be built around git repos and if I need to work disconnected, or work with command line tools, I just clone it onto my laptop and merge back later. And being the old “curmudgeon” C/C++ embedded tools guy, that’s a significant admission…

IDE in the Cloud

There’s been a lot of talk at EclipseCon here about the “IDE in the Cloud”. I missed it but apparently the Mozilla Bespin demo at the e4 talk was quite impressive. It is easy to set up, and it’s pretty fast as a code editor. I guess that makes sense since browsers have been heavily optimized for fast display of textual and graphical content. So I have to admit, it’s led me to reconsider this technology. Rational tried this many years ago, remember Catapulse?, to build hosted development environments. The idea, and the company, collapsed in the high-tech bust of 2001. Maybe they were just too early.

But as we start to rely on the cloud, what happens when the sky’s are clear? I just heard Kevin McGuire say that, hey, my e-mail is on the server, why shouldn’t my code be there too. I use Outlook and Thunderbird in a mode that downloads my e-mail to my machine so I can access it disconnected. I don’t always have access to the cloud. Until the cloud is omni-present, through things like metropolitan wireless schemes, and cheep, $10/day at my hotel wasn’t cheep, then I don’t see it really taking off as the typical way you do software development.

I just worry that the whole “Cloud” paradigm is being pushed by the people that can profit most from it, not from an inherit need of potential users.

CDT Code Introspection APIs

I’m just sitting in Markus’s talk on the APIs the CDT provides to create code models, ASTs, the Indexer and ways of getting information out of there. These facilities are used to implement features such as content assist and open declaration and searching, etc. And you can use it too for all your static analysis needs. A copy of his presentation is available from the EclipseCon site by following the link to gPublication. It’s a great reference on how to get started.

http://www.eclipsecon.org/2009/sessions?id=685

Slides from my Talk on the CDT Community

I’m not sure why the EclipseCon submission system isn’t accepting this but here are the slides from my talk on the Rise and Fall and Rise of the CDT, Lessons on Building Communities. I had a lot of fun creating and presenting this talk. I guess it’s a passion of mine and people seemed to like it. The slides have few words on it and I’d be happy to give more details if you have any questions. Feel free to add comments here.

A Great EclipseCon Already

I’ve been here at EclipseCon for 48 hours and it’s been great already. I’ve met a lot of people here, like I always do, and it’s a good sign that vendors still see Eclipse as important, enough to spend the money in these tough times to send their technical experts.

EclipseCon is an important conference, especially for those who are looking to get started with Eclipse and to grow their expertise. There’s probably not enough time to learn it all, but at least you know where to go look when you get home.

It’s also important for projects to get the word out about the great work their doing and to grow their community. I was pretty disappointed about the number of submissions I had for the C/C++ category. But we’re making due. And it’s good to see the number of DSDP contributors here giving talks so we’re still showing the world that Eclipse is more than just a Java IDE.

Other things to report:
– work on ObjectiveC is starting and the guys managed to get a prototype running based on CDT late last night.
– the Ada project is looking to reboot.
– e4 talks are the hottest ticket in town
– adding git continues to be controversial and we really need to get a bigger ground swell of contributors who care about making this happen

Anyway, beer time. More later.

POWERVR goes MP

I was just reading up on the news that Imagination Technologies has launched a new generation of their architecture that drives POWERVR. What’s POWERVR, you ask? It’s a good question, but chances are, if you have a mobile device that has 3D accelerated graphics, it’s driven by this hugely popular silicon IP.

The big news is that even they’re going multi-core to achieve scalable graphics. They claim they rival the performance of discrete graphics chipsets, which I assume means the nVidias and ATI’s of the world. That’s a pretty interesting combination when you look at the latest chips that have multi-core ARM processors combined with DSPs (digital signal processors) for audio/video processing, which can then be combined with these powerful 3D cores.

I can imagine some really powerful handheld and other mobile devices based on these things. I just wonder about power consumption, the eternal challenge for mobile developers. We’ll see how that plays out, but the power of these things also blurs the gap between mobile and x86 based “netbooks” and even set-top boxen. Interesting times ahead, indeed.

But as good as the hardware is, you still need the software to make them come to life. It’s going to be an exciting time for developers that target this market. I just wonder if there are enough of them. This is one area where I’m hoping the Eclipse DSDP and CDT projects can help. We need an easy to setup package to give the students and hobbyists, the future commercial developers for these platforms, the tools they need to get them started. This is something I’d be happy to talk about at EclipseCon and see if we can get a community to start putting this together.

Should Wascana get Qt?

I’ve been working on my EclipseCon talk and tutorial, which I’m really looking forward to now. Especially the talk where I hope to share some of the things I learned being involved with the CDT project for almost seven years now. And I’m hopeful it’s useful to others working in open source, especially the things that didn’t work as well as I hoped…

Anyway, while doing that, I’ve started playing with Qt, which was recently released with LGPL as one of the choices for licensing. That eliminates one of the hurdles I have for including it into Wascana, my CDT for Windows distribution. So now the question is, should I include it? Should I also keep wxWidgets? Since the Wascana plan is to become a p2-based distribution, there should be no harm in having both. It’s just that I don’t use wxWidgets so I’m not sure whether what I’m producing works or not.

Why that matters is because I build the libraries for Wascana myself using the gcc/g++ I distribute with it. I’ve switched over to tdragon.net as my supplier for gcc, mainly because they (or he?) provides the latest releases from gcc (and the mingw.org community is a bit of a mess). And I want to make sure these libraries are built with the latest and greatest optimization algorithms gcc is providing.

At any rate, I’m learning how to build Qt and it’s running right now. I hope it doesn’t melt my laptop. They have a ton of example and demo projects that I can use to test so it’s a pretty good environment for producing quality distros.

I should also look at their Eclipse plug-ins. They provide an installer for the Windows version which installs into an existing Eclipse. I wish they had a p2 repo that I could use instead. But I’ll have to see what they’re laying down and understand their licensing to see whether I can make these available in Wascana as well. I still find it weird they have their own IDE, Qt Creator, when it should be pretty easy to put together an Eclipse-based IDE that does almost the same, if not more…

Emulation versus Multi-Platform

Over the last little while, when I could steal away a few minutes from my busy day job as of late, I’ve been thinking more about open platforms for handheld and set top consoles that makes good use of the 3D hardware that is becoming common place in these platforms. Of course, being open, I’m talking about open source and royalty free APIs like Linux and OpenGL and OpenGL ES.

I’ve been very excited about the LGPL’ing of the Qt C++ application framework. The programming paradigm is very clean and the library set is huge, including my favorite, the WebKit browser, and, of course, OpenGL support.

One angle I’ve been following is my other favorite piece of open source software, the qemu CPU emulator. The community is very active there and they’re are more and more platforms and hardware components being emulated making it a nice platform to play with some of the concepts I’m thinking of.

But the one critical piece missing is 3D hardware emulation. I would be great if someone could put that together, and given the comments on my previous blog entries, it would be very popular. I could do it, but I have the bigger picture in mind and have some other things I’d like to work on, in particular, a port of the Clutter concepts to Qt (and no, the QGraphicsView is close but not quite what I was thinking of).

Thinking about it tonight, it struck me. Since Qt is multi-platform and the tools I use, gcc and the CDT are multi-platform too, why don’t I just start this work on Windows using Wascana, the Eclipse CDT/mingw gcc package I am also trying to work on. I could even use the OpenGL ES emulation libraries that are out there to make sure these ideas work with OpenGL ES, another multi-platform component. And as my work here progresses, I will have the opportunity to use the same tools to make it work on real hardware.

It’s this type of environment that motivates my work on the CDT and what makes it so powerful. It’s time to put it to work for me.

Qemu 0.10.0 now available

In a message from Anthony Liguori to the qemu-devel mailing list, “The QEMU team is pleased to announce the availability of the 0.10.0 release. This release has been a year in the making and is the result of almost 3,000 changesets from around 80 developers.” The release was quickly put together in the last few days (i.e. in Eclipse terms, there wasn’t much of a ramp down). And the resulting build is actually pretty hard to find, at least until it gets propagated to all the mirrors, but it’s a milestone anyway.

There’s a pretty long list of new targets and hardware emulation. There’s also improvements to the VNC support. The biggest news is the new code generator for translating op codes into runnable code. It’s called TCG (Tiny Code Generator according to Wikipedia) and removes the dependency qemu had on gcc 3.x, which means it gets to take advantage of the huge performance improvements of gcc 4.x. I think I read somewhere that the Android qemu was running 1.5 times faster thanks to that. And my copy is humming using the gcc 4 I’m integrating into the future Wascana 1.0.

This release is a good sign that the qemu developers are getting serious about formal releases. It’s been quite a while since the last release and there has been huge architectural change. That has led to a some difficulties co-ordinating with the other projects that use it, like the kvm Linux virtualization platform. Having quality releases more often will make it easier to get changes from forks into the mainline, which is good for everyone.

For me, it’s just good fun to be able to have a virtual platform to try different ideas with C/C++ libraries that target mobile devices that I could easily target to real hardware some day. And I think it’s intriguing the Eclipse integrations we could do to make it easier to work in that environment. A lot we have already, like CDT (of course) and the Remote System Explorer for launching, and there is work on more coming from the Tools for mobile Linux project. Interesting indeed.

Qt 4.5 Full of Surprises

Qt 4.5 was released today. As expected, it comes with the adoption of the LGPL license which should make Qt free for commercial development if you so choose. I think this will be a huge boost to the adaptability of Qt and hopefully make an impact in the Gnome versus KDE Linux desktop wars (mind you the KDE gang are really shot themselves in the foot when they released 4.x as a planned regression from 3.x).

There were a couple of big surprises that came with the release announcement. First was the discontinuation of Qt’s embedded product, formerly and lovingly known as Qtopia, the best product name in the business ;). The plan I understand is to fold that functionality into the Qt mainline releases with the intention of making Qt a truely cross platform API for both desktop and mobile. And they’ve started by introducing an OpenGL ES paint engine into this release, which if I understand correctly will allow you to use Qt to create apps for mobile devices now.

The other big surprise, isn’t really a surprise since I’ve known about and wondered what was going to happen, but they’ve released a 1.0 version of their IDE, Qt Creator. It appears they are going forward with this IDE as a key offering, and have released it as LGPL as well. Which means, theoretically, a community could form behind it and it could start to compete with Eclipse. We’ll need to do a thorough analysis and understand what it all means. Especially since they also uprev’ed their Eclipse integration and have a beta of their Visual Studio integration available. How does Qt Creator fit into this world?

The only thing that really bugged me in their news release on Qt Creator is their claim that “it provides the first IDE designed specifically for cross-platform development”. Uh, guys. The CDT is 7 years old and counting…