Saturday, August 18, 2007
Wednesday, April 11, 2007
But let's take our eyes off the future and focus on the here and now. While we are dreaming (or should I say 'provisioning') new Update dreams, the old Update still needs to work and tie us over, or the new Update will be crushed by the unrealistic expectations before the birth pangs subside. There is Europa to ship, new features to post, patches to publish for all that brilliant software that was bug-free when you unleashed it on the unsuspecting public. Unfortunately, Install/Update is down to one active committer (yours truly), and he is a manager with a full plate. That's as if we are down to 0.1 committer with a short attention span :-).
As I am typing this, Alex Blewitt ran away with the prize by blogging about this first. I am not going to out-Alex him but here is the plan for a few good men (or women - we are equal opportunity here :-):
- Take a look at the Update inbox (using this query will help)
- Pick a bug that you would really like to see fixed and that has enough steps to be reproducible (but call it first so that others don't investigate the same problem)
- Pick a fairly recent build (M6+) and set it up for Update development using these instructions (read under 'Feature-based self-hosting')
- Try to pinpoint the problem; if you feel confident, try to fix it
- If it seems to work, post a patch to the bug report, but be diligent - we want to fix existing problems, not create the new ones, right?
What's in it for me, you ask? You can make your code a proud part of Eclipse Platform (names and emails of contributors will be prominently displayed in file copyright notices). You can get a high from fixing a hard problem. At the end of 3.3, we will create an 'Update Hall of Fame" with pictures and short bios of top Update contributors. Finally, you will get to help yourselves by fixing problems that affect your own projects and components (he is looking at you, Mylar :-).
Last but not least, you will learn the whole problem domain of installing and updating bundles. When the time comes to switch to the new and shiny Equinox Provisioning, you will know what works, what doesn't, what you like and what is, to use Ed's immortal words, sucky.
Thursday, February 22, 2007
API Usage Assumption: Every aspect of the API matters to some Client.
This quote comes from a seminal document written by Eclipse veteran Jim Des Rivieres who was in his time known as 'the API Police'. He struck fear in many a component lead when the time came to inspect the APIs before the freeze. Jim is now busy doing other interesting things (although he still lovingly prepares Eclipse New&Noteworthy for the milestone builds), but I encourage every committer to frequently come back to the said document, particularly when she prepares for the major API surgery (note from the revision history that the document has been kept fresh, most recently updated to include rules related to JDK 1.5).
Read the document and you will learn how to walk the razor-thin line between the API bliss and the M5a.
Wednesday, February 21, 2007
How can one reach such a lofty goal? It takes years of experience, ever increasing responsibilities that compete with the proper testing time and most importantly, a blissful disregard for the Eclipse API prime directive: Every API matters to some client.
In M5, we have reworked the UI Forms. A color that was used in the past is no longer needed. I deprecated the color key but made a mistake of not leaving the color by that key around for those who want to use it. You can read all the embarrassing details in the bug 174441. Who knew people find the color Forms use to paint the section title gradient useful for other interesting purposes.
What can I do to top this? I have never caused a rebuild of a GA build. But something tells me my commit rights will be wrestled away from me before I manage to achieve it.
Friday, February 16, 2007
Here's what happened to me the other day: I was happily working in my basement office when my visiting mother-in-law decided to call all the people in her phone book, one by one, from the main floor. She did it from a wireless handset that works on 2.4MHz. My wireless access point works on the same frequency, and in the battle that ensued the access point lost. Without the connectivity, I decided that I cannot do anything useful and chose to continue after lunch, or after my mother-in-law reaches the page 'Z', whichever comes first.
Beside the potential 'Everybody loves Raymond' moment, where am I going with this? The point is that I was not able to do anything useful without the network. This issue came to the head a few days ago when the UA committer Curtis D'Entremont and myself discussed the new remote help feature in Eclipse. It allows you to put all your help content on a remote server and still have the same great experience without the megabytes of documentation on your hard disk. We were trying to work through the scenario of remote help users that lost connectivity.
We came up with some great technical ideas but in the midst of it I remembered the mother-in-law incident. Should we even bother? Can you do anything useful without the connectivity any more? Have we crossed the networking Rubicon?
Tuesday, February 13, 2007
Today I visited Steve, the daddy of the SWT team and he gave me a demo of Eclipse on Vista. The buttons that shimmer, the text fields that glow with anticipation when the mouse moves closer, the tree view animations, the works. It all looked great, fully native and had great performance.
I expected to see a number of entries in M5 New & Noteworthy to properly celebrate this event. Instead, this is what I found:
Note that the Win32 port of SWT continues to work well on Windows platforms and fully exploits the new look and feel of Windows Vista.
Have we come to assume that SWT 'just works' for so long that we don't see a great achievement when it smacks us on the head? I for one will follow Wassim's suggestion and say the following:
Thank you, Steve and the SWT team!
Friday, February 9, 2007
The title of the blog and the matching header I painstakingly selected from a list of templates should reflect that I care about pixels. My several prior lives as a developer in IBM somehow always revolved around the UIs. I am the proud founding father of JFace before it was made part of Eclipse, and one of the original members of the Platform UI team (and I have an 'Eclipse Founder' shirt to prove it). I am also the father of PDE (as you can see, I am big on parenting), but all the damage I have done is now systematically rectified by the current PDE owner Wassim Melhem. As Scott Adams would say, since then I have been moved to an area where I can do the least damage - the management. The only sandbox I can play in to earn my developer stripes are now the UI Forms.
UI Forms started in PDE and many people still think they are a part of it. I don't blame them, since I used PDE as a show case for the technology. However, UI Forms are now completely standalone and are sitting much deeper in the stack - as an optional RCP plug-in. To me, they have it all - all the pixels I can eat, the excitement, the sexy (if you don't believe me, believe the blogger Chris Aniszczyk and one of his recent posts). And this is a great time to be in Forms, being recently refreshed and with all the new goodies that you can read about in the Eclipse 3.3 M5 New & Noteworthy (at the time of writing, M5 has not been posted yet, but my pent up creative juices cannot wait on the Eclipse process to play itself out).
And now, the $5,000,000 question: why should I add this blog into my bulging feed aggregator? Because: 1) you like Eclipse, 2) you like sexy UIs and 3) you understand that UI Forms can get you there if only you could ask the guy who wrote that code how the darn thing works.
And who knows, maybe, just maybe, I may be able to share a pearl of wisdom beyond the pixels every once in a while. They say old people are wise. That's a lie and you know it but I am willing to play along.