August 2005 Archives

Well, the migration of mail from CIS to CNS is done. Our IMAP is handled by them. Our SMTP is handled by them. Our authenticated, off-campus SMTP is handled through them (via the as-yet-undocumented server). Our Webmail is handled by them. Mailing lists are handled by them through the LISTSERV.

I've setup MX records for the important hosts and working on setting up the rest. We've reconfigured Pine and Mutt on the local system to use them. (Pine presents a nice view of both their folders and old CIS local folders. Mutt has always offered the ability to read any folder you hand, so getting local is just "c ~/Maildir/"). I've updated the documentation on how to setup email so that everyone can set up their own clients. And I've spammed most of the faculty, staff, students a few times to let them know. Speaking of, I probably need to spam the students again...

Most of my readers will probably consider the above to be self evident, but I'm mostly ranting about a specific bug I've encountered with the Google Maps API map I've put together for my church web site, newhopeks.org.

Try viewing that page in MSIE (5.5+ please, since that's what Google supports) and see what you get. (Well, it might be fixed by the time you read this, but if not...) If you're lucky, it'll just work. If you're not lucky, you got a box saying:

Internet Explorer cannot open the Internet site http://www.newhopeks.org/map

Operation aborted.

If you're especially unlucky, when you clicked OK, it took you to a page saying "The page cannot be displayed." (This error comes up even though you could see the map and everything up until you clicked OK.)

So, what's the problem? It turns out that if you have some JavaScript (the language Google Maps is written in) and you use a technology called DOM to manipulate the page (which Google Maps does) and you put the script in a <script/> tag inside of a <div/> tag or a table, MSIE freaks out some large percentage of time. I haven't even gotten the error consistently even on the same machine. This is complete idiocy. There's no way an error like this should have made it into production software and if it did, it should have been found and patched ages ago. This is what regression testing is for. Especially since this is a very well known error (just google for ""Internet Explorer cannot open the Internet site" "Operation aborted"" if you don't believe me).

The moral of the story is, don't use Internet Explorer. I recommend Firefox or Apple Safari or Opera or any of the Mozilla-based alternatives (except for Netscape as AO-Hell has ruined it).

Oh, well. Being the web developer who has to make his stuff work everywhere, I now I have to figure out how to move my script out of all the <div/> tags that make up the New Hope theme...

Well, this isn't much of a brag, but I have had the opportunity to play with the new Google Maps API. We've had some nifty map ideas for church and I've started putting together a map for us.

Right now, all it shows is three markers: one for where we meet (Flint Hills Christian School), one for our church office on Poyntz, and one for our undeveloped land on Scenic Avenue. Not very exciting. However, I plan to eventually add a way of marking addresses on the map so that when there's an event or announcement with an address, that can be made into a link that's then geocoded on the map. There will also be a special map to show where various small groups meet with popup links taking you to each group's summary page, pictures, etc.

As I said, not much of a brag, but it's surprisingly easy to play with (only 5 lines of JS code and 2 lines of HTML and you're off running) and pretty fun for being JavaScript.

Okay, I've been using SF.net for a long time for the Contentment project for almost four years to the day (started August 5, 2001). Well, today I said good-bye. I'm not requesting a project removal or anything that drastic. Think of this as more of a separated married couple rather than a divorce, but I'm contemplating the divorce.

Anyway, the event that actually brought it on was the fact that I was having a hell of a time trying to get Brian D. Foy's extra-handy release script to login to SF.net. I still don't know what was wrong. I dug for about an hour and began to evaluate the value SF.net was adding to my project and decided I was getting too emotional about it. SF.net wasn't providing me an hour's worth of effort in value.

What do I need in project management? Well, I need a source code repository. I've quit using CVS for the most part in preference of Subversion. However, SF.net has stated plans for, but not an implementation of Subversion, yet. So, I'm not using any of the public CVS facilities.

I need a way to make source tarballs and release them for download, preferrably on a mirror, in case my project ever gets popular. As a Perl project, the most natural choice for this is CPAN anyway. Uploads to CPAN require a two-step process that is so easily automated (in about 30 lines of code after the FTP upload), it's laughable to see that the same process on SF.net takes over 200 lines of code (again, after the FTP upload). CPAN also offers anyone who wants to give Contentment a shot the ability to just install it like:

cpan Contentment

that will download the latest version, unpack, build it, test it, and install it. SF.net can't do that (nor should it, really), but again, I don't really need them as another mirror. (Btw, at this time, this isn't the best possible way to install it, but it does work. I hope to make this a fine way to install eventually...)

I need a web site. Well, I need a web site I can control to a greater extent than SF.net offers. I have hosting through Dreamhost and can even use FastCGI to make database connections persistent. SF.net doesn't offer any such thing (of course, I pay for DreamHost). In any case, I don't use SF.net's web hosting, database, etc.

I need to post news. Well, heck, this blog can do that and I don't have to use the pathetic little "News" thing SF.net offers. No. No way.

I need to have bug posts. The "tracker" on SF.net is not what I consider bug tracking. It's more like a headache. My favorite bug tracker is RT. Handily enough, CPAN offers an RT bug tracker.

What else might they have? Hmm... donations. Well, that's easy enough to setup through PayPal directly. Why do I need them for this? PayPal also skims less off the top. I'm not getting any donations anyway. Mailing lists are also nice, but I don't really have anyone to talk to just now and I can get a list through Perl.org and have my project identified with Perl directly or I can setup my own mailing lists with DreamHost.

The SF.net web site is ad-ridden, slow, and doesn't really provide anything I can't either provide myself (Subversion repository and public mirror) or can get through CPAN (most the rest). Anyway, I'll keep my project alive there for now, but unless SF.net makes some drastic improvements, I imagine I'll ask for project removal.

SF.net was a neat idea 4 years ago, but the improvements that have been made haven't really impressed me that much. They've been very incremental and not very drastic. The paid side of things sounded like a great idea when they announced it, but they offer things like a better search tool and more ways to get project news. Bah. Who cares? I want better project management tools. If you gave me some compelling features, sure, but not for what they offer.

It seems to me that SF.net needs some innovators to come in and revamp the place. It doesn't even look that different than it did 4 years ago. Stagnate. They should fix that. If they do, perhaps I'll come back. In the meantime, I'm headed over CPAN where real changes have taken place in the last 4 years.