Blogging Roller

I think number of units sold is a big factor in the <a href= "http://www.rollerweblogger.org/page/roller/20021010#dave_on_software"> Dave on software curve. If you have sold hundreds of thousands or millions of units of your software then you have the resources to fix lots of bugs, do usability studies, write a really nice installer, etc. You also have the motivation to do do things right the first time because one irritating bug can cause a tremendous number of tech support calls.

I was not really happy with the quality of my writing on last night's <a href= "http://www.rollerweblogger.org/page/roller/20021010#dave_on_software"> Dave on software post. My writing sounds crazy and rambling. I actually pulled the story a couple of times to try to fix it. Then I realized two things. Firstly, the story had already been cached by my reader's RSS aggregators. And secondly, like it or not, "crazy and rambling" is my own personal style and I shouldn't try so hard to hide that.

Continuing on the quality theme: Rebelutionary noticed a Roller bug. Roller's RSS syndicator does not expand macros in weblog entries. That is now issue <a href= "http://opensource.atlassian.com/projects/roller/ViewIssue.jspa?id=10110">ROL-64.

[with <a href= "http://jakarta.apache.org/commons/sandbox/sql/">Commons SQL] generating other stuff from your XML file should be trivial; like the Castor mapping file or beans to map 1-1 to your relational model. Just use the commons-sql beans with Velocity or Jelly etc. [James Strachan]

Now I see how Commons SQL could be useful in Roller. We need a better way to generated DDL for varous databases and we need a better way to generate Castor JDO mappings and data objects as well. We've been <a href= "http://sourceforge.net/mailarchive/forum.php?thread_id=1134736&forum_id=9297"> talking about this for a while, maybe Commons-SQL is part of the answer.

I'm going to go out on a limb here and post for all to see my recently developed theory of commercial software quality and value vs. software price. Three disclaimers are necessary. First: I am a heads-down, propeller-head, bits-and-bytes, grunt-coder, who reads Slashdot for news - so what do I know. Second: my theory is based mostly on hearsay and second hand knowledge. And third: I believe that pretty much all software sucks. So, saying that PhotoShop or Dreamweaver "rocks" is somewhat disingenuous on my part. Anyhow, please bear with me and take a look at my whiteboard:

The basic idea here is that software quality and value goes up as the price of one installation of the software goes up until you get to around the $3K to $5K price range. After that, quality and value start to drop off and around $20K to $50K they fall right off the cliff.

From my graph you can see that I consider consumer software like Quicken, Office, and PhotoShop to be pretty good stuff, with good value and few major flaws. You can continue to get better quality and value as you get into the AutoCAD and Microsoft Visual Studio Enterprise price levels, but after that quality and value start to drop off. Once you get up around the $100K to $1M level you are in the painful world of systems integrators, e-Commerce consultants, and enterprise software. At this level, usability features that you take for granted in consumer products are often missing, installation instructions can run into the hundreds of pages, and teams of consultants are needed to do almost any customizations.

As a J2EE developer who works in the field of enterprise software, I want to do better - I want to learn how to make software suck less. But, enough about me. Let's talk about you. What do you think? Is there any truth to my theory? Am I totally wrong about this? Is there any insight here, or am I just being petty? Does the ROI of the big expensive enterprise software make it the most valuable of all, quality be damned?

Matt is picking up my slack and making bunch of usability and UI enhancements to Roller 0.9.6-rc1. He has done the most work in this area of any of the Roller crew and he the best looking Roller based weblog around - based on his x2 theme which is now part of the Roller distribution.

There have only been four downloads of Roller 0.9.6-rc1 so far and I can understand that. Who'd want to download and install Roller only to have to do it again next week. Still, we are getting some good bug reports and finding things that should be fixed before the final 0.9.6. I don't want to release 0.9.6 until our two big customers are happy with it. I'm hoping to get some time this weekend to wrap up the User Guide, fix some RSS output issues, and wrap up the pluggable authentication stuff. Hopefully, 0.9.6 final will be the weekend after that.

I just bought a Dell 450mhz PC with 256mb RAM, 8gb disk, 17 inch monitor, keyboard, track-mouse, 2 power-strips, and a UPS for $100. What a deal! I'm working on setting it up for my wife tonight instead of working on Roller.

Matt Raible has installed RC1 and is finding and fixing some small issues. Hopefully Anthony Eden will be trying RC1 tonight. In other news... Anythony reported a problem with Roller 0.9.4's RSS output that still needs to be fixed.

Okay, I'm too tired right now to rant. But I'm going to. And when I do it's going to be long. It's going to be hard. It's going to be a rant like you've never seen a rant before in your life.[Russell Beattie, posting at a temporary URL ]

If I don't take a break from Roller development tonight, my wife will kill me. Oh sh*t, she saw me. Gotta run...

This is a release candidate. The Installation Guide has been updated, but the User Guide remains to be updated for the final 0.9.6 release. You can download Roller 0.9.6-rc1 at SourceForge (thanks VA Software!). Make your contribution to Roller by reporting any problems that you find to Roller's <a href= "http://opensource.atlassian.com/projects/roller/BrowseProject.jspa"> JIRA issue tracker (thanks <a href= "http://www.atlassian.com">Atlassian!). And finally, thanks to all the Roller contributors and supporters!

UPDATE: Here are some more convenient download links:

roller-0.9.6-rc1.tar.gz - All you need to run Roller (on Tomcat)
roller-src-0.9.6-rc1.tgz - Full source code, build scripts, etc.
roller-tools-0.9.6-rc1.tgz - Jars needed to build, extract into source dir

Summary of new features:

   Multi-user :
   - Admin UI for deleting users
   - Better start page: pages through users, shows time of last update
   - Allow users to change their email addresses and passwords
   - Plugin authenticator
   
   Weblog editing enhancements:
   - Ekit HTML editor applet is available as an option to users
   - Option to save weblog entries without publishing them
   - Better control over weblog entry publish date
   
   New and improve macros:
   - Protection from recursion in includePage and showWeblogEntries macros   
   - Big archive calendar macro shows titles for each day
   - Expand/collapse feature in Bookmark and Newsfeed macros
   - Users can control number of weblog entries displayed
   - New macros, including those necessary for RSS templating
   
   RSS related features:
   - Multiple RSS feeds for each weblog, one for each category
   - RSS feeds available with full-text or excerpts only
   - RSS aggregation features may now be disabled
   - RSS syndication output is now cached

I think 0.9.6-rc1 looks pretty good, so I will probably update the install guide, and release it tomorrow night. Later in the week I'll try to do some stress tests, some memory leak tests, and wrap up the user guide for a final 0.9.6 release.

Roller now supports all the right macros for RSS templating. For example, I was able to take Mark Pilgrim's RSS 2.0 template for Movable Type and convert to a Roller template to get RSS 2.0 output from Roller.

All systems go! Now it's time to add some of the new features to my templates and to do a little tweaking. I'll start my tweakings with the new big archive calendar.

A little later... Now that Roller allows you to put macros in weblog entries, mentioning the macros.showWeblogEntries() macro in a weblog entry as I did on <a href= "http://www.rollerweblogger.org/page/roller/20020927#macros_to_support_rss_templating"> 9/27/2002 is the recursive kiss of death. I think I can use a thread local storage trick to prevent this and the macros.include() page recursion. Guess that means RC2. Correction: guess that means I have some work to do before I actually release RC1.

Here is a cool new one: macros.showRSSLinks() displays links for the various types of RSS 0.9.1 feeds available from your weblog. Roller now gives you two feeds per category, one with full text and one with excerpts. You can even invoke it in a weblog entry, see:

$macros.showRSSLinks()

Ok, so the formatting is not that nice and it does not get expanded in the RSS feed. Hmmm...

Russell Beattie has had a tough week, but he is OK - at least in the physical world. In the digital world of cyberspace, however, he is not doing so well. First, he lost his client by accidentally installing Linux over his Windows partition. Next, he lost his server because his ISP, CWIHosting.com, has mysteriously shut down his account. CWIHosting tells him this is because of "police reasons." The CWIHosting support people told him that he needs to email the CTO and CEO to get any further information. Unfortunately, they are not responding to his emails. He is a little worried that he might not be allowed to get into his account and rescue his weblog archives. That is a scary thought.

Russell thinks that "police reasons" might be actually be a mis-spelling of "policy reasons" and perhaps he simply overloaded his shared Java VM by misconfiguring something when he set up OSCache. I hope that is the case. Anyway, Russell is setting up a new account at JohnCompanies.com ISP and hopes to be back on line by Monday or Tuesday. Good luck Russell!

Russell Beattie's site has been down for a day or so. I hope everything is OK. Anybody know what is going on?

Last minute features and fixes are underway for Roller 0.9.6. Stay tuned!

OSCache does in fact rock. And it rocks even if you are not using JSPs, as I have mentioned before. Roller pages don't use JSP and they are still able to take advantage of OSCache goodness.

The Mozilla Composite Editor works pretty well, but it seems to eat some spaces every once andawhile (see what I mean) . I have also had intermittent problems getting it's link editor to work. The editor is only at version 0.0.5,so I have confidence that it will be fixed and will eventually become my editor of choice.

I've been using the Ekit editor, but it has some problems as well. The main problem with Ekit is the Java Plugin. The Java Plugin is just not that stable, at least not on Mozilla 1.1. Every third or forth time I hit the post button to submit a weblog edit, Mozilla locks up! I wonder if this is a JRE 1.3.1/WinXP/Mozilla specific issue. Has anybody else(hi Matt and  Russell) experienced such flakeyness on other platforms?

With this idea in place, we would single-handedly and instantly capture the entire blogging market [...] Husband and wife Ben and Mena Trott, creators of the perl-based Movable Type, are going to be like deer in headlights when Skribe moves Movable Type over into the junkyard! [a post from the Skribe dev forum]

That is some tough talk for a project whose CVS archive is empty.

Wow, that looks cool! I'm editing this post with it now. Very nice.