REVERT TO CONSOLE » Jeff

Setting Up Trac on Windows

Jeff — Wed, 02 May 2007 15:13:52 +0000

This is an adaption of this howto, without all the options. I’m setting up a test Trac project without SVN, and this is how I did it:

Install Python
Install Trac
Install ClearSilver
Install PySQLLite
Install setupTools. Apparently you can just double click this, but I ran it as a python script from the console so I could see what’s going on
I just set up the most rudimentary authentication system as described here:
1. Use this python script to generate a file:
  from optparse import OptionParser import md5 # build the options usage = "usage: %prog [options]" parser = OptionParser(usage=usage) parser.add_option("-u", "--username",action="store", dest="username", type = "string", help="the username for whom to generate a password") parser.add_option("-p", "--password",action="store", dest="password", type = "string", help="the password to use") (options, args) = parser.parse_args() # check options if (options.username is None) or (options.password is None): parser.error("You must supply both the username and password")
  # Generate the string to enter into the htdigest file realm = 'trac' kd = lambda x: md5.md5(':'.join(x)).hexdigest() print ':'.join((options.username, realm, kd([options.username, realm, options.password])))
2. Execute it like this:
  python trac-digest.py -u username -p password >> c:\digest.txt
3. Now use trac admin and add this user and admin permission you created the file for:
  python trac-admin.py /path/to/trac/project permission add username TRAC_ADMIN
To install the handy webmin for a project, first make sure you have created a user with admin rights (as described in the above step), download the egg.zip from here, rename from *.egg.zip to *.egg, and place it in the project/plugin directory. Then restart trac.
To install the timing and estimation plugin, download it from here. This comes as a zipped source file. To generate the egg, unzip the archive, cd to ./trunk, and find the setup.py file. Run
python setup.py bdist_egg. this should create, among other things, a dist directory. The egg should be located there. Copy the egg to the /project/plugin directory. This constitutes installation step one on the website that they were too lazy to write.
Be sure that the plugin is enabled. Add “timingandestimationplugin.* = enabled” to trac.ini (in the [components] subheading)
Run $trac-admin /path/to/projenv upgrade

* You should get a message like this:

Timing and Estimation needs an upgrade
Upgrading Database
Creating bill_date table
Creating report_version table
Upgrading fields
Upgrading reports
Upgrading usermanual
Done Upgrading
Upgrade done.
Restart Trac and you should be ready to go

Trac vs. XPlanner Death Match

Jeff — Wed, 21 Mar 2007 16:08:08 +0000

Trac kills XPlanner.
Trac has more functionality. Trac has plugins and is easy to customize. Trac provides extensive reporting capabilities out of the box. Trac is more intuitive.
The only thing XPlanner has going for it is that it makes use of buzz words… actually just a single buzz word: Story.

And the UI is so much better in Trac!

Bad:

Good:

Mad Command Line Skillz

Jeff — Fri, 02 Mar 2007 13:26:54 +0000

I’ve been working with shell scripts recently. The *NIX command line has some really powerful tools, but I’m afraid to say I’m starting to feel pretty comfortable in my Java IDE land, or in Python land, hell even in Ruby land. And going back to shell scripting has not been fun for me.

A few tips:

Use the debugger. Add -x to the shebang line and get this:
#!/bin/sh -x
Send output to a file:
ls -l > ls-l.txt use this all the time so you don’t have to cut and paste from the terminal.
Send stderr to a file too Why stop with stdout?
grep da * 2> grep-errors.txt
Merge them Go crazy, why have two files when you can just have one?
grep da * 1>&2
Go both ways, no, I’m not promoting both sides of the fence, but
grep * 2>&1
Run in Stealth Mode Don’t try this at home.
rm -f $(find / -name core) &> /dev/output.txt

Managed Processes Post 2

Jeff — Wed, 21 Feb 2007 20:28:26 +0000

This is a follow up article to First Steps Toward a More Managed Process I wrote at the end of last month.
I’ve been using Trac with the Timing and Estimation plugin now for about a month. I tried to promote it to my team for our next product but they never considered it and instead we’re using XPlanner. XPlanner is still a 100% improvement over the tool we used before, which was called the nothing whatsoever tool.

In my opinion Trac does everything that XPlanner does plus includes about 10 more useful features… such as Subversion integration, Bugzilla integration, dozens of useful reports, exporting out of the box to comma-delimited and tab-delimited text files, ical integration, rss feeds… plus Trac can be deployed as a standalone app (as I have done on my desktop), or integrated with a number of different backends and frontends. XPlanner uses MySQL. I’m really beginning to think Java is far behind in web development, and the community is growing fatter and wider (re: hibernate+spring+whatever bloated front-end) rather than slimming down and actually innovating like Python & PHP (simple extensible frameworks)… but I digress… I’m really just here to talk about the first month of implementing my own managed process.

In short, Trac has helped me manage a list, but I still haven’t reached the managed process nirvana I strive for.

What I’ve done with Trac:

I’ve separated my work projects into different Trac components. I am currently working on 3 different components. One is in QA, another is being maintained, and a 3rd is new development. So I’ve been successful at managing these components and managing my tasks for each.
I’ve mapped the Iterations to Trac Milestones. This is really slick because Trac provides a nice RoadMap and Timeline reports for work by Milestone.
Each task I call one of the three out of the box Trac types; either a defect, enhancement, or task.
The only piece I haven’t really figured out the best way to use is the Trac version. Since where I’m currently working milestones/iterations are also versions, it’s not like I could have a version 2 of iteration 3. I tried to coordinate a version with a use case code, but that hasn’t turned out to be effective.
I made use of Trac priorities, which create nice color-coded reports.
I didn’t make enough use of the timing and estimation plugin. I need to do that next month.

Going forward:

Going forward I’m moving to a monthly milestone. Since XPlanner will track iterations, I’d like to use Trac to manage my own tasks and velocity. I see this as a shift away from a project planner, to more of a personal planner. I’d like to track my tasks, my own velocity on a project, and keep notes for each independent of what the project planning app is doing.

Wiki Formatting Issues

Jeff — Mon, 12 Feb 2007 17:20:23 +0000

I’ve had some wiki formatting issues lately, particularly when I try to add unix commands like scp and wget. The *NIX Command Line Tips section has suffered as a result.

But no more.

I’ve finally discovered a work around. If you’re adding something to the wiki page and you see a red underline under a particular word, and when you go to save the page you get an error message, try wrapping the offending word in
tags and see if that helps.

Executable Jars with One-Jar

Jeff — Thu, 01 Feb 2007 01:34:29 +0000

I’ve been working on packaging a batch process so it can be run by some other group… nothing’s better than passing off the work onto someone else… (gotta love consulting!)

In doing so I’ve stumbled upon the problem of packaging and delivering a command-line app. This is where One-Jar project comes in. The promise of One-Jar is that you can package everything together and run it from anywhere. This is nice if, for example, you want to pawn off a batch process app to a support group that doesn’t know how to package and deploy stuff.

This post shows how I went about doing that, and the resulting ant target I created, as well as some of the pitfalls I experienced along the way.

The batch process app I put together is really just a single class that has a ton of dependent jar files, as well as some config files I want to override. When everything is set up the way it’s supposed by set up via One-Jar, the resulting exploded structure looks something like this:

/applicationContext.xml
/log4j.properties
/hib1.hbm.xml
/hib2.hbm.xml
/main/main.jar – my single class file packaged here * also contains a manifest that points to my main
/lib/ – a bunch of dependent jars
/com/* – the one-jar classes
/META-INF/MANIFEST – this manifest points to the one-jar classloader

To make a one-jar you need to do the following:

Create your own main jar file with a manifest that points to your main method. In my case my manifest looks like this (notice I’m not worrying about the classpath):
Manifest-Version: 1.0 Ant-Version: Apache Ant 1.6.5 Created-By: 1.4.2_04-b05 (Sun Microsystems Inc.) Main-Class: com.batch.stuff.MyMainClass
Create your executable jar that contains a lib directory with your dependent jars, and your main directory and main jar as above. Also include your property and config files at the base like they normally would be. *My friend Steve pointed out that it’s probably best to test the jar at this point. If you’ve created the jar you should be able to run the jar from the same directory…
And make sure you don’t accidentally add the one-jar jar into the jar you’re packaging.
Update your executable with the exploded contents of the one-jar jar. This is easily performed via the command-line, like so:
$ mkdir boot $ cd boot $ jar -xvf ../one-jar-boot.jar $ jar -uvfm ../myapp.jar boot-manifest.mf .

After working with this for a while, I was able to create an ant task that can do it.
Here’s the ant task I created to jar up my batch process:

depends="init, compile.myapp" description="Jars the contents of the classes directory into a myapp specific jar">


	
	

	

		

			

			

		

	
	

	

	
	        basedir="${myapp.basedir}/classes"

		update="true"

		manifest="${myapp.basedir}/classes/myapp-manifest.mf"/>
	

	
		update="true">

		

			

		

		

			

		

		

			

			

		

	
	

	
		dest="${myapp.basedir}/boot" />
	

	
		update="true"

		duplicate="preserve"

		manifest="${myapp.basedir}/boot/boot-manifest.mf"

		basedir="${myapp.basedir}/boot"

		/>

This creates an executable jar file that can be executed like this from any directory on any system:

java -jar MyApp.jar

Alternatively, to see the classloader info:

java -Done-jar.info -jar MyApp.jar

Other Resources

This is a good article
written by the author of One-Jar.

Supposedly you can use an eclipse plugin too…

First Steps Toward a More Managed Process

Jeff — Fri, 26 Jan 2007 22:34:30 +0000

I hate to say I’m moving toward an agile process and risk becoming too buzzword-focused. But I’m quickly realizing how important it is to have an effective methodology for managing yourself as a consultant. I’ve used task lists for quite a while as a developer, and that’s fine for the most part.
But as a consultant, you really have an extra layer of self-management. I’m always being asked what I’m working on, when it will be done, etc. I started my own custom task list in Outlook and shared it with the team lead, but I’m thinking that it’s really a narrow solution. In addition I’m often seen as outside the regular flow of management; and by that I mean sometimes I’m viewed with more suspicion since I’m new and a consultant. I’m also more likely to be given the side tasks not the core tasks, and sometimes those are overlooked. In my particular case, the management of our project is so bad, they’re having trouble even coming up with a task list or agreeing upon features for a given iteration.
That’s what brought me to thinking of how I might go about implementing my own agile methodology. Given that requirements are never fixed here, tasks are not well defined, and iteration deadlines are the only thing that are noticed, I need to find a way to get myself there. And the more detailed information I can give my team lead about my tasks, the issues that come up with each task, and estimates for completing them, the better.
Also there’s the issue of what happens if something goes bad and the customer gets angry? Again, the more detailed information I can provide to show what I’ve been working on and when, the better.
Today I started down that path… setting up a locally running TRAC instance to manage my own tasks.
I’m hoping this will help when I’m assigned my next major task, such as the scheduling app that just went to QA.

Holy Grail of CSS Layouts

Jeff — Tue, 23 Jan 2007 15:04:56 +0000

Jon Udell referenced this article yesterday in his blog, about the holy grail of the 3 column CSS layout.

It’s amazing how illusive standards support is across browsers, despite the ubiquitous nature of the web today.

Agile Anyone?

Jeff — Thu, 04 Jan 2007 03:04:05 +0000

I’ve been reading up on eXtreme programming practices. I’ve really been in a world of hurt at my current project. The engineering team is adequate, the technology I understand, but the management and requirements are a nightmare. I’m thinking that if I don’t act I’ll really find myself in trouble; the use cases are a disaster, my direct manager is so scattered she doesn’t even respond to my emails, and the higher up managers are non-existent.
I think this is an extreme situation and I should really take matters into my own hands. Particularly, if I can’t manage my own requirements gathering and management, as well as my own time, I’m going to be in trouble.
So I ask the site, does anyone know of any good agile methodology resources? I’m looking for a way to implement this on my own desk… please comment if you know of any good websites.

Assembla Me Something Interesting

Jeff — Wed, 27 Dec 2006 19:07:02 +0000

I found this site while trolling the blogosphere today: Assembla looks like a sourceforge kind of site, only with newer web tools. Looks like it provides Subversion, Trac, and other collaborative features…