I spent a great deal of time today doing more hacking on Paladin, a trend I've had the last week or so when I wasn't working on getting my book out the door. The Code Library feature, for now, has been withdrawn until I can get it working well, which it just doesn't right now -- everything works the right way except for synchronization, which is the entire point behind the feature. The code for it is still there, but it won't be enabled for the next release. I fixed several really annoying bugs, including workarounds for two bugs in Haiku which took a lot of effort to find but were trivial to work around once found. One of them is project files being mistakenly identified as plain text files.
Source control is the star feature for the next release. I intend to make it about as easy as possible to integrate source control into a developer's workflow as possible. It took me a long time to really "get it," but now that I do, I want others to have an easy time using it without having to learn command line syntax until they're ready to do so. The only thing that I currently have on the to-do list before the release is to polish and test the different ways that source control is used. Once that's done, I'll be putting out a release. I really love this project -- the rewards are incredibly satisfying. :)
Wednesday, July 28, 2010
Thursday, July 22, 2010
Learning to Program With Haiku Now Available in Book Format
The book is finally done! Getting through the proof copy took *so* much longer than I ever expected. Luckily, right now I'm out of town with a lot more time on my hands, so I had a lot more time to be able to sit down and get through it. It has been published through Lulu.com so that a great deal more of the profit from the book goes to me instead of the pockets of a book retailer. Here is the link to the book and e-book.
Learning to Program with Haiku at Lulu.com
The regular price is $25 USD, but as part of the launch of the book, it is on sale for 15% off the regular price through August 15 when you use the coupon code BEACHREAD305 at Lulu. It will also be available through online book retailers like Amazon.com, Barnes and Noble, and others, but it will be a couple of months before it appears on those sites.
Learning to Program with Haiku at Lulu.com
The regular price is $25 USD, but as part of the launch of the book, it is on sale for 15% off the regular price through August 15 when you use the coupon code BEACHREAD305 at Lulu. It will also be available through online book retailers like Amazon.com, Barnes and Noble, and others, but it will be a couple of months before it appears on those sites.
If you've been a fan of the lessons and want to show appreciation or to have a copy of it on your desk while you work your way through, now you have a chance to have a high quality copy, and if you've been sitting on the fence about it, read a lesson or two and then decide for yourself. Enjoy!
Thursday, July 15, 2010
Paladin: No More CVS, Either
I wish that when I started the project for Paladin over at Sourceforge that I had been able to choose Subversion for source control using the client for Zeta. Because of an outdated client, I was forced to use CVS. While CVS is (barely) adequate for legacy work, anyone who knows the importance of source control knows better than to use it when far better tools exist. Subversion is CVS done right, but it certainly isn't perfect, either.
Distributed source control tools are the current fashion, it seems, and rightly so. After a lengthy discussion of Mercurial vs git which to choose over Subversion, I decided to do some looking into both of them. It's another one of those Coke vs Pepsi, KDE vs GNOME kind of controversies. Long story short: I haven't looked into git very much, but I've learned enough about Mercurial to say that I like it a lot.
Just a few moments ago I checked in the current development snapshot to Paladin's new Mercurial tree. It's official: I'm no longer using Zeta for development. I am still supporting source compatibility for the platform, but now my focus for Paladin is solely Haiku. This change in perspective will bring in some features not possible for R5 or Zeta.
One new feature for Paladin 1.3 is integrated source control on a basic level. The inspiration comes from The Joel Test: 12 Steps to Better Code. One of the steps is using source control -- yes, even CVS. In all the years that I've been flinging C++, I never used source control locally. I didn't see the point in doing so locally and didn't think all that much about source control to begin with. I didn't want to put in the effort to learn more, either. Lazy developer, I know. git and Mercurial make it foolish to not use it. It's a different workflow, but only slightly.
Paladin's integrated source control will be simple and about as unobtrusive as possible. It creates a repository when you make a new project. Click a menu item or press a shortcut on the keyboard, type in the check in message, and it will check in your work locally. Click another menu item and it will push your changes to another repository somewhere. It's about as painless as it can get. Mess something up? Revert to the last known good revision. You will be able to even choose from Mercurial, git, and Subversion for your work. Only the basics will be covered in 1.3 to avoid pushing the release off too much. Gotta get back to hacking. TTFN
Distributed source control tools are the current fashion, it seems, and rightly so. After a lengthy discussion of Mercurial vs git which to choose over Subversion, I decided to do some looking into both of them. It's another one of those Coke vs Pepsi, KDE vs GNOME kind of controversies. Long story short: I haven't looked into git very much, but I've learned enough about Mercurial to say that I like it a lot.
Just a few moments ago I checked in the current development snapshot to Paladin's new Mercurial tree. It's official: I'm no longer using Zeta for development. I am still supporting source compatibility for the platform, but now my focus for Paladin is solely Haiku. This change in perspective will bring in some features not possible for R5 or Zeta.
One new feature for Paladin 1.3 is integrated source control on a basic level. The inspiration comes from The Joel Test: 12 Steps to Better Code. One of the steps is using source control -- yes, even CVS. In all the years that I've been flinging C++, I never used source control locally. I didn't see the point in doing so locally and didn't think all that much about source control to begin with. I didn't want to put in the effort to learn more, either. Lazy developer, I know. git and Mercurial make it foolish to not use it. It's a different workflow, but only slightly.
Paladin's integrated source control will be simple and about as unobtrusive as possible. It creates a repository when you make a new project. Click a menu item or press a shortcut on the keyboard, type in the check in message, and it will check in your work locally. Click another menu item and it will push your changes to another repository somewhere. It's about as painless as it can get. Mess something up? Revert to the last known good revision. You will be able to even choose from Mercurial, git, and Subversion for your work. Only the basics will be covered in 1.3 to avoid pushing the release off too much. Gotta get back to hacking. TTFN
Saturday, July 10, 2010
No More Hangs for Paladin
Those of us who run Haiku under a multicore machine can finally breathe a sigh of relief. For about as long as I can remember -- probably as long as it's had multithreaded builds -- Paladin has suffered from problems with locking up under Haiku. There never was a problem with it when I was developing Paladin from within R5 or Zeta, though.
Over the last couple of days, I've been doing some serious development on Paladin from within Haiku and one product of said hacking has been Haiku bug #6288. I've been working pretty closely with Ingo Weinhold on the problem and he was very kind in teaching me a few system-level tricks along the way, particularly with the kernel debugger. The result? Revision 37460 patches the problem. Considering that several bugs that Paladin development has exposed are still open AFAIK, I'm really thankful for Ingo patching this one -- I wouldn't have had a clue where to begin.
All of this effort I've been putting into Paladin this week has resulted in some new features and bugfixes. Running a project in the debugger now works again following a patch to both Terminal and Paladin, for example. I'm going to finish one major feature, see if I can fix problems in the code library, and start going into the test phase for a new release. I haven't a clue when that release will be, but, worst case, it should be here by the time I have to go back to school next month, barring major problems -- hopefully much, much sooner. Until next time. :)
Thursday, July 8, 2010
Not Worried About Haiku's Future
Operating systems are funny things. They take FOREVER to develop, for example, and I don't just mean Haiku. Look at the development time for Vista. It took a multi-billion dollar corporation for than five years to write the upgrades for XP.
There's also a chicken-and-egg problem that fledgeling OSes have: how to get an OS without applications and how to get applications without an OS. Luckily for Haiku, a lot of that problem has been mitigated by BeOS' startup history. The QT4 port just saw a 4.7 release. I'm really thankful for the efforts of those guys. Other people like Michael de Vincius Oliviera and the HaikuPorts team have been hacking away at porting stuff from other platforms. While they certainly don't have all of the BeOS goodness of a native app, at least we have some of these apps and games. KOffice takes a team of dedicated developers that is IIRC larger than our OS team alone. We're really starting to get a lot of useful apps.
Stability has always been my sticking point for coding under Haiku, but at least in my experience of late, Haiku's been pretty stable, all things considered. I'm typing this from a GCC4 hybrid build that I installed from sources. So far Paladin hasn't been plagued by deadlocks like it has been in the past, which has led me to believe that it was never Paladin's problem in the first place, but the jury's still out on that one. I'm even working on more features that will only work under Haiku and fixes for the platform, like using rsync to properly synchronize modules for the code library. In the last couple of days I worked with a couple guys on the team to nail down a bug in Terminal which was preventing Paladin from running applications in the debugger. After patching both Paladin and Terminal, it works beautifully now.
Feature-wise, Haiku is even now doing an admirable job of playing catch-up. People in the past have mentioned that R2 would be the release where Haiku would implement all the features that BeOS was missing that all the other OSes have, but more and more I don't think it'll take that long. OpenGL stuff might, but the Google Summer of Code has been good for that. There have been other people who couldn't wait that long for stuff like WiFi, so even now that partially works. Simple amazing, if you ask me. Then again, even though it's been slow, Haiku's progress from the outset has been that way, too.
There's also a chicken-and-egg problem that fledgeling OSes have: how to get an OS without applications and how to get applications without an OS. Luckily for Haiku, a lot of that problem has been mitigated by BeOS' startup history. The QT4 port just saw a 4.7 release. I'm really thankful for the efforts of those guys. Other people like Michael de Vincius Oliviera and the HaikuPorts team have been hacking away at porting stuff from other platforms. While they certainly don't have all of the BeOS goodness of a native app, at least we have some of these apps and games. KOffice takes a team of dedicated developers that is IIRC larger than our OS team alone. We're really starting to get a lot of useful apps.
Stability has always been my sticking point for coding under Haiku, but at least in my experience of late, Haiku's been pretty stable, all things considered. I'm typing this from a GCC4 hybrid build that I installed from sources. So far Paladin hasn't been plagued by deadlocks like it has been in the past, which has led me to believe that it was never Paladin's problem in the first place, but the jury's still out on that one. I'm even working on more features that will only work under Haiku and fixes for the platform, like using rsync to properly synchronize modules for the code library. In the last couple of days I worked with a couple guys on the team to nail down a bug in Terminal which was preventing Paladin from running applications in the debugger. After patching both Paladin and Terminal, it works beautifully now.
Feature-wise, Haiku is even now doing an admirable job of playing catch-up. People in the past have mentioned that R2 would be the release where Haiku would implement all the features that BeOS was missing that all the other OSes have, but more and more I don't think it'll take that long. OpenGL stuff might, but the Google Summer of Code has been good for that. There have been other people who couldn't wait that long for stuff like WiFi, so even now that partially works. Simple amazing, if you ask me. Then again, even though it's been slow, Haiku's progress from the outset has been that way, too.
Subscribe to:
Posts (Atom)