Dev-Picayune

I had always admired Konfabulator but after trying it out, I stopped using it because I didn’t want to pay. Instead I had switched to Rainlendar and Rainmeter since they were free and offered some of the same coolness. But now that Yahoo has purchased Konfabulator and made it free. I think I’ll have to check it out. It will be interesting to see a comparison of memory and CPU usage between the two. So stand by for more news on my future comparison. It’s my understanding that Konfabulator really offers the closest thing to the Apple Quartz environment (in terms of creating widgets), so let the fun begin.

As far as I know, I now have a working trackback system in place for this site (Dev-Picayune). As I noted previously, the biggest challenge was handling PHP coming from the XSL transformation. Once I got that working, the rest was just the normal tedious details. At the same time, you may notice I did a little cleanup on the header graphics to make them more consistent and a little more 3D accurate. I also used my own modification of the ALA Sliding Doors method for dealing with variable sized backgrounds under text (so my “trackbacks” bubble sizes itself). So now I can use the same graphic for a slightly differently sized “comments” bubble or a tiny bubble that just says “b” or a giant bubble with the words “when the revolution comes, the Perl programmers will be the first ones against the wall”. Well, maybe that won’t quite fit… but it certainly might be a true statement.

Apparently, when using Python 2.4 and py2exe, the executables created by this newer version of py2exe are dependent on msvcr71.dll — a Microsoft created .dll that some feel is not entirely distributable without a Visual Studio license. Luckily, I have a Visual Studio license, so I was able to distribute it. Looking at my machine, I also noticed that Office 2003, MS SQL DE, and a few other pieces of software are apt to install this, so chances are moderately good that someone will already have this on their machine. But I do find it unfortunate that Python can’t be compiled more independent of Microsoft related stuff.

I found myself in a quandry in the midst of my trackback implementation. A normal implementation in blog software involves listing the total number of total trackbacks on the listing page. Then, for the full entry page or a separate trackbacks page, the actual details of the trackbacks are displayed. The issue I had was that I was using PHP to perform the XSL transformation on my XML blog contents file. The trackbacks are controlled by a funky Perl script (that I have since modified). I wanted to be able to do a PHP server-side include bit of code inside the results of the XSL transformation. Luckily, looking at this article on evaluating php expressions using Sablotron, I discovered how to get PHP to render the result and then evaluate the inner PHP expressions that I included in the XSL.

Key to this whole thing is using:

<xsl:processing-instruction name="php"&gt php stuff goes here... </xsl:processing-instruction>

Next, you need to change the processing slightly in the php code to look like this:

$xh = xslt_create();$result = eval ("?>". xslt_process($xh, './devblog/current.xml', 'devblog.xsl'));echo $result;xslt_free($xh);

And you’ll be good to go. So that takes care of the count so that I can show the number of trackbacks for each entry and it will also provide me the ability to include the trackback details inline with the single entry display page.

After looking into doing TrackBack support, I decided to also look into providing comment capability. However, after just a short amount of research, I have realized that Blogger won’t support comments unless you have post pages turned on. In Blogger.com’s world that means that you get a separate HTML page for each post. Since I don’t use that “feature” and instead generate my post pages dynamically with XSL from an XML file, it would be a total waste of time and space to have a duplicate of what I already have. Duplicates that I am afraid would never be used since the blogger template does allow putting the comments in the main generation. But they are not generated unless you have post pages turned on. There are some third-party solutions I’ll look into, but I am not so sure I want to bother with that. It’s bad enough being dependent on one third-party to store my blog (BTW, many blogger definazis would say that I don’t have a blog at all since I don’t support comments).

It isn’t like I use DasBlog, but it would be cool to create a theme that other people actually use. If only I had more time. I could install DasBlog on a local machine and design a theme for it. Apparently, Scott Hanselman is calling for DasBlog themes/theme designers.

I am looking into implementing TrackBack for both posting and posts on my blogs. The unfortunate thing about using Blogger is that you have to go third-party to piece together stuff like TrackBacks. I know I should be doing comments first but TrackBack’s get me exposure on other people’s sites as well. So really my first step is to try to implement sending TrackBack pings to posts that I reference.

Thankfully, Blogger finally has a final solution to the layout problems they were causing with their new image feature stuff. The solution involves a new blog setting under the format tab. No more needless div’s. While I am relieved that it is solved, and while it has also encouraged me to improve my XSL and general markup for the better, I see another lesson in software development here. What could have been done to prevent this problem?

As I see it, there were at least 2 or 3 other possible methods for resolving the problem or not causing it in the first place. They could have updated the standard set of layout templates with the necessary additional markup embedded in the template. This would force nearly everyone to upgrade to the latest version of the template since most have at least some minor template modifications. But for those not using the new image feature, it would pose no problems.

Another and better method would be to modify the CSS for each of the standard templates with a clear:both line in the leading and trailing elements for each post. This is obviously time consuming as each of the available templates would have to be modified and those running older templates might not get the changes. I believe this would have worked, but again, probably would have required a lot of up front work. But then you end up with less superfluous markup on the page itself and special templates should be unaffected.

Yet another option would have been to add a new blogger template tag that doesn’t include the new additional div’s and immediately notify everyone what they were for. Then folks like myself could switch to the new tag without a serious hickup. Alternatively, the existing tag really should have stayed the same, and a new tag created that would include the new div’s. But again, that would require folks changing over to a new template.

My thought here (rather than some grand solution) is just that someone obviously believed that including two new div’s at the beginning and end of every post would not pose a serious problem. I don’t understand why they believed that exactly, but certainly it was some amount of short sightedness. It’s a problem that either a little more design time up front, or some more thorough testing could have avoided.

It seems that I was not the only one affected by Blogger’s changes. GirlHacker’s Random Log was affected by Blogger’s image feature as well. At least I have my fix in place now (at least on 2 of the 3 blogs). Again, it’s nice to at least have most of the control on my end of things. The fix involved my CSS file and my XSL file, but nothing else. But getting back to Lily’s blog. In my mind, she’s like one of the blogging pioneers. She’s been doing a lot longer than most.