Unruly Rambling - cvs tag

Spring Cleaning delivers a needed tutorial for OS X users.

Sun, 08 Apr 2007 20:50:01 GMT

A month or so ago, I got the keen idea that I wanted to replace a 19" boat anchor of a monitor with something flat panel to gain some desk space. This altruistic goal was accomplished after a 20" Samsung wide screen display showed up on my doorstep for less than $200. I was happy since I reclaimed a nice big corner of my desk. Even though I accomplished my original goal, the printer sitting on the end of the desk irked me too. Unlike the monitor, I'm not able to replace my printer with a smaller model. At least not one that would be significantly smaller. The best I could do was find a better location where it wasn't taking up a square foot of desktop space. My decision was to put it in the desk where two linux boxes resided. One of the linux boxes runs this site, a personal wiki, and cvs repository. The other sits mostly idle as a large backup file server. In order to maintain my current setup, I would need a new smaller computer to do these lightweight tasks. Mac Mini to the rescue! I picked up a refurbished Mac Mini from Apple for less than $500. It's a duo core 1.5ghz which is probably 8 times faster than the linux box it replaced. The good news is that it is also 10 times smaller and looks fantastic sitting out in the open. In addition to relocating the printer in the tower compartment, I also moved the spare paper, DSL modem, and various other office items. My desk looks fantastic now. Don't quit reading yet, there is a point to this entry and it has nothing to do with the neatness of my desk.

Instead of installing CVS on the new Mac Mini, I decided that I would migrate to subversion. This was not as easy of a task as I thought it would be. As I sought to find a decent tutorial online, I came up short time and time again. I never did end up finding anything that I could follow step by step and get to where I had a subversion repository front ended by Apache, using Apache authentication running on OS X. Instead I used bits and pieces of content from a half dozen sites to get everything to work. For those of you that are in search of something similar or have run into this same brick wall, the following tutorial is for you. I did my best to document every step along the way. If you run into something that is ambiguous or needs updating, please let me know. I intend to keep this up to date.

HOWTO: Installing Subversion(1.4.3) and Apache(2.0.59) on OS X(10.4.9)

CVS obfuscation

Sat, 17 Dec 2005 18:38:42 GMT

A couple days back, I had a really weird thing happen while updating the source code of a project while using IDEA 5(Windows XP). When I would issue the cvs "check directory status" command, it would tell me that the class in a particular package had been removed. After seeing this and knowing that I didn't remove that class, I issued the command again. Now I see that the update put the files back in my workspace like they were new. This went on for a couple hours, acting flakey from time to time. By the end of the day I recreated my workspace and checked everything out from scratch again but the same problem continued to happen. I assumed it was my workspace because our automated build process never had a hiccup. Eventually I went to the cvs repository(Linux) directly to poke around. Low and behold, I saw 2 directories, one using camel case and one using an all lower case naming convention. After looking even further, I noticed that one of the directories was empty but had an Attic directory. The other directory had the mysterious files. Upon inspecting the Attic directory, I saw that files by the same name had been removed from the repository.

In other words, the directory structure looked like this
com/acme/mypackage/Attic/WeirdFile.java,v
com/acme/myPackage/WeirdFile.java,v

My guess is that this weird behavior was happening because I was working on windows where nothing is case sensitive and it couldn't differentiate between these two directories. Bottom line, if you must develop software on inferior operating systems pay attention. Refactoring could cause you to be in this very situation scratching your head.

NoFluffJustStuff Gateway Java Symposium - Day 1

Sat, 20 Mar 2004 04:41:35 GMT

For the second year in a row, I have been fortunate enough for my company to spring for admission to the NoFluffJustStuff Gateway Java Symposium. Well today was the first day of the 3 day conference. Registration began at noon and the introduction started at 1pm. The day was comprised of 3 seminars, dinner, and finally the expert panel discussion where the panel answered questions from the audience. The thing that I really like about these kinds of conference's is the close interaction with the speakers. One thing that I was a bit disappointed with was the fact that there was no keynote address like last year. Dave Thomas was the keynote presenter last year and did a fantastic job. Oh well, maybe next year. Below, i've listed the presentations that I attended and my thoughts on them.

Decoupling Patterns - untangling that knot of code (Dave Thomas)
This was probably the most frightening presentation that I've seen to date. It's like Dave has been working on my team for the past three years. In detail, Dave described situations that my team faces on a weekly basis. He also touched on several key indicators that your code might be becoming too coupled and ways to remedy the situation. In a nutshell, he recommended package level visibility to across the board with one or more facade's into the package. This seems like a reasonable approach and was fairly straight forward. The thing he didn't spend much time on was the return type's or objects from the facade. This is the part that I have always struggled with. If package a depends on package b, what package should the shared objects(return types and parameters) belong to? This seems quite easy if they are domain objects, but 95% of the time this is not the case. In the past i've come to the conclusion that duplication of code(i.e. data classes not business logic) may sometimes be a good idea if it makes dependency management easier. All in all, this presentation was everything I expected it to be.

Advanced Version Control with CVS (Dave Thomas)
To start off, this presentation wasn't quite as "Advanced" as I would have liked it to be. It dealt with more of the day to day operations that a configuration manager would deal with. I did learn a few tricks along the way though. The merging of changes between two tags into head was something I didn't know about and seemed very useful. I have every confidence that I could have went to the resident CM guy at work(not by choice) and he would have guided me in the right direction. The things I wanted to learn about were advanced ways to hook code formatters up cvs so all commits were formatted appropriately. I have been an end user in the past were this happened but I don't have the foggiest idea how to set it up. I would have also liked to learn more about the various triggers that can be set up when commits occur. To make a long story short, the pieces that I hoped for would not have applied to most CM processes. They are unique things that Im interested in. Obviously, Dave's talk wasn't catered to me :)
Dave, as always, was a tremendous speaker keeping the attendees attentive at all times. I really enjoy the injection of his British humor in each and every facet of technology.

Classloading in Java: Building Dynamic Systems without pain (Stuart Halloway)
I missed Stuart's presentations last year so I didn't really know what to expect. He started off with, "This isn't one of the cool new topics or buzzwords that most people came to hear about. Classloading hasn't drastically changed since the Java 1.2 days....". His witt was just as entertaining as the topic he was covering. With that being said, this presentation alone will likely save me hours and hours debugging classloader issues that were somewhat foreign to me in the past. Like most Java guys I understood what a classloader did, the hierarchy of classloaders, heck I even screwed up my own classpath a time or two. The things that I hadn't thought about or read about were the gems here. Learning the differences between implicit classloading, explicit classloading, and mixing the two was invaluable. The best part was that he wrote code on the fly to demonstrate both. I learn best using these hello world typle applications as building blocks and then expanding on them. I will try and attend more of his presentations throughout the weekend.

Expert Panel
David Geary, Mike Clark, Stuart Halloway, Dave Thomas, Erik Hatcher
The expert panel was pretty good. Topics discussed included off shoring, next new big thing, thoughts on JDK 1.5, JavaServerFaces, and dynamically typed languages. All in all, the discussions were good. Audience participation wasn't the best, but that might have had to do with the fact that is was 8:00pm on a Friday night. Maybe tomorrow's will be different, who knows.

No More Windoze

Sun, 01 Feb 2004 04:04:31 GMT

Well, the time has come to find a new distribution since Redhat will no longer be providing support for 8.0(current version supporting this site). Since Im most familiar with Redhat, Fedora was an obvious choice for my new distro. To ease into this transition and guarantee zero downtime, Ive decided to install Fedora onto the only Windoze box that I currently have. Thats right, my only windoze box will run windoze no more. I guess one could argue that Im upgrading the OS. My plan is to install everything on this machine that my current Redhat 8.0 server provides. My current "can't live without" installed apps are JBoss, MySQL, JAMES, and CVS. Once these are all up and running, Ill do the same to the Redhat 8.0. This will give me poor mans failover in case of a vaporized hard disk or some other hardware failure. I say poor mans failover since redirecting traffic will involve manual reconfiguration of my router.

Fedora Core
Bootup with Disk 1 in CDROM
1) Language Selection
English
2) Keyboard Configuration
U.S. English
3) Mouse Configuration
Generic Wheel Mouse (PS/2)
4) Monitor Configuration
Sony Multiscan 17sf
5) Installation Type
Workstation
6) Disk Partitioning Setup
Manually partition with Disk Druid
7) Disk Setup
/dev/hda1 nfts
/dev/hda2 /boot
/dev/hda3 /
/dev/hda4 swap
8) Boot Loader Configuration
WinXP /dev/hda1 default
Fedora Core /dev/hda3
9) Network Configuration
hostname: pegasus.shoesobjects.com
manual network settings
10) Firewall configuration
No firewall
11) Additional Language Support
none
12) Timezone Selection
America/Chicago
13) Set Root Password
13) Package Installation Defaults
customize software packages to be installed
+ KDE Desktop Environment
+ Server Configuration Tools

Fedora HOWTO apt-get Download the apt-get RPM for Fedora Core here. Once you have installed it, you will need to retrieve an up to date package list using the following command apt-get update. Now that we have the latest package listing for everything installed on the system, we should upgrade the installed packages to the latest and greatest versions. Upon running apt-get upgrade, I see that I have 32 packages to upgrade. By typing a Y at the prompt, I will begin to download all of the upgraded packages, 56.3MB to be specific. Gee, this kinda feels like windows update except I don't have to reboot afterwards :)

Other Packages Installed

More to come.........