My new project is going well so far. It's nice to bill for the commute and
work from home twice a week. I'll be finished with the prototype for our first
project today, and we'll start pair-programming on Thursday to implement it.
Since one of my main objectives is to teach my supervisor everything I know, I
think the best way to do this is to pair-program. I tried giving a couple of
presentations on Hibernate and XDoclet, but my supervisor is a Java Greenhorn,
so even that was a little advanced.
Our first project is to automate job posting to the various job posting boards
the client subscribes to. Currently, a person in HR enters the job onto their
website, and then visits each job site (or sends an e-mail) and manually enters
the information. The application we're developing will allow them to enter the
job on their website, select which boards they want to post to, and submit the
form. I initially received a list of 32 websites and newspapers. After
contacting them all, I'm amazed the archaic systems that are in place for this
"automation." Most sites will accept a simple e-mail in whatever format we
want. In my mind, this means a person is going to manually read the job
posting and manually enter it into their system. Great, it's easy enough for me
to compose an e-mail. 2 out of 32 allow for FTPing text files in a certain
format, and 1 site has an XML format (not DTD, just a format) that they expect
to be attached to an e-mail. This is a *huge* opportunity for webservices and
a common XML Schema (if you ask me).
The one thing that's been slightly frustrating is getting a development
environment setup. I (as usual) need to setup CVS, a bug tracking system,
automated tests, etc. It's a Windows shop, so no Linux allowed. Not a big
deal, but I haven't been able to get VPN access yet. No VPN means that I can't
checkin stuff from home - which is dangerous IMO. My supervisor suggested we setup everything on my Linux
box at home, which is fine with me, but could be a lot of maintenance on my
part (permissions, UPS, backups). For bug tracking, I'm leaning towards
Bugzilla over Scarab because I'm familiar with it. I'd like to use JIRA, but
don't want to shell out $800 to impress a client with some nifty bug tracking
software. Free is always easier to sell to clients. I've actually thought
about buying a JIRA license for Raible Designs and hosting my clients bugs
(while I'm on the project), but it might be a pain to transition once I leave
(I suppose I could rent the space to them or something). So what I'm asking
for is - if you had to administer your own Linux box for your day job - how
would you do it? Groups? Backups? UPS? If I don't do it on this project, I'll probably
do it someday.
