Well, it's official: My IT Manager convinced those in the upper echelons (well, considering it's a non-profit with only around 20 employees, that meant the president and the CFO) that (1) he and I need to attend a PHP conference, (2) due to the amount of work we've been putting in to bring money into the organization, cost shouldn't be too much of a deciding factor, and (3) php|Tropics isn't too expensive, especially considering the sessions involved cover some of the very issues we've been struggling with the past few months (PHP/MySQL/Apache and clusters, PHP5 OOP, PHP Security, test-driven development, Smarty, and more).
So, we're going to Cancun in May!
This is incredibly exciting! I've never been to Mexico, nor even a resort, so I'll finally get to find out what my wife and friends have been talking about all these years. Plus, the conference is top-notch — many of the presenters are well-known in the PHP community, and have blogs I've been following for the past year. (I only wish that Chris Shiflett's PHP Security series wasn't running head-to-head with the PHP5 OOP Extensions and PHP 5 Patterns sessions; I suspect Rob and I will have to do a divide-and-conquer that day.)
Drop me a line if you'll be attending — I'm looking forward to meeting other PHP junkies!
I've been extremely busy at work, and will continue to be through the end of March. I realized this past week that I'd set a goal of having a SourceForge website up and running for Cgiapp by the end of January — and it's now mid-February. Originally, I was going to backport some of my libraries from PHP5 to PHP4 so I could do so… and I think that was beginning to daunt me a little.
Fortunately, I ran across a quick-and-dirty content management solution yesterday called Gunther. It does templating in Smarty, and uses a wiki-esque syntax for markup — though page editing is limited to admin users only (something I was looking for). I decided to try it out, and within an hour or so had a working site ready to upload.
Cgiapp's new site can be found at cgiapp.sourceforge.net.
Shortly after I wrote this original post, I figured out what the strength of Gunther was — and why I no longer needed it. Gunther was basically taking content entered from a form and then inserting that content (after some processing for wiki-like syntax) into a Smarty template. Which meant that I could do the same thing with Cgiapp and Text_Wiki. Within an hour, I wrote an application module in Cgiapp that did just that, and am proud to say that the Cgiapp website is 100% Cgiapp.
1.5.3 fixes an issue introduced by 1.5.2 that creates a performance hit whenever the run mode is being determined by function name or CGI parameter. More details on the Cgiapp download page.
At work, we've been developing a new platform for our website, based entirely on Cgiapp. This week we released the first stage of it: garden.org and assoc.garden.org. These should stand as good testament to Cgiapp's robustness!
With all that development, and also with some communication from other Cgiapp users, I've made some changes to Cgiapp, and release version 1.5.2 this evening.
1.5.2 is mainly security and bugfixes. Error handling was somewhat broken in
1.5.1 — it wouldn't restore the original error handler gracefully. This is now
corrected. Additionally, I've made
run() use the array returned by
consisting of the
$_POST arrays — in determining the run mode.
Finally, I've modified the behaviour of how
run() determines the current run
mode: if the mode parameter is a method or function name, it cannot be a
Cgiapp method or a PHP internal function. This allows more flexibility on
the part of the programmer in determining the mode param — words like 'run' and
'do' can now be used without causing massive problems (using 'run' would cause a
race condition in the past).
As usual, Cgiapp is available in the downloads area. Grab your tarball today!
I picked up on this article on Friday, glanced through it and moved on, but noticed this evening it had been slashdotted — at which point I realized the author is the current CGI::Application maintainer, so I looked again.
At my first glance through it, it appeared the author was looking for a nice, easy-to-use pre-processor script for generating a site out of templates and content files. To that end, he, in the end, recommended ttree, part of the Template Toolkit distribution.
However, the real gist of the article — something that should probably have been summarized at the end — is that the author was looking for a free and OSS replacement for DreamWeaver's Templates functionality. This functionality allows a developer to create a template with placeholders for content, lock it, and then create pages that have the bits and pieces of content. Finally, the developer compiles the site — creating final HTML pages out of the content merged with the templates.
Now, I can see something like this being useful. I've used webmake for a couple of projects, and, obviously, utilize PHP in many places as a templating language. However, several comments on Slashdot also gave some pause. The tone of these comments was to the effect of, "real developers shouldn't use DW; they should understand HTML and code it directly." Part of me felt this was elitist — the web is such an egalitarian medium that there should be few barriers to entry. However, the webmaster in me — the professional who gets paid each pay period and makes a living off the web — also agreed with this substantially.
I've worked — both professionally and as a freelancer — with individuals who use and rely on DW. The problem I see with the tool and others of its breed is precisely their empowerment of people. Let me explain.
I really do feel anybody should be able to have a presence on the 'net. However, HTML is a fragile language: trivial errors can cause tremendous changes in how a page is rendered — and even crash browsers on occasion. The problem I see is that DW and other GUI webpage applications create, from my viewpoint, garbage HTML. I cannot tell you how many pages generated by these applications that I've had to clean up and reformat. They spuriously add tags, often around empty content, that are simply unnecessary.
The problem is compounded when individuals have neither time nor inclination to learn HTML, but continue using the tool to create pages. They get a false sense of accomplishment — that can be quickly followed by a very real sense of incompetence when the page inexplicably breaks due to an edit they've made — especially when the content is part of a larger framework that includes other files. Of course, as a web professional, I get paid to fix such mistakes. But I feel that this does everybody a disservice — the individual/company has basically paid twice for the presentation of content — once to the person generating it, a second time to me to fix the errors.
This is a big part of the reason why I've been leaning more and more heavily on database-driven web applications. Content then goes into the database, and contains minimal — if any — markup. It is then injected into templates, which go through a formal review process, as well as through the W3C validators, to prevent display problems. This puts everybody in a position of strength: the editor generating content, the designer creating the look-and-feel, and the programmer developing the methods for mapping content with the templates.
There's still a part of me that struggles with what I perceive as an elitist position. However, there's another part of me that has struggled heavily with the multitasking demands made on all web professionals — we're expected to be editors, graphic designers, programmers, and more. In most cases, we're lucky if we're strong in one or two such areas, much less passionate about staying abreast of the changing face of our medium.
It's about 50 minutes shy of 2005 here, and Maeve has finally succumbed to sleep, I'm almost done with my stout, and we're in West Bolton without TV for the second year running (yay!).
I hope the new year brings peace and happiness to one and all! Happy coding!
I don't know why I never bothered to look this up, but I didn't. One thing I
typically do in my parent Cgiapp classes is to pass
to the template. I just found out — through the pear-general newsgroup — that
this is unnecessary: use
$smarty.server.KEY_NAME to access any
your template might need.
It's now been confirmed: I'm a geek.
Okay, so that probably comes as no shocker to those of you who know me, but it's the little things that make me realize it myself.
I've been frequenting Perl Monks for a couple of years now, mainly to garner ideas and code to help me with my personal or work projects. I rarely post comments, and I've only once submitted a question to the site. However, I do frequent the site regularly, and the few comments I've put in — generally regarding usage of CGI::Application — have been typically well-moderated.
Well, yesterday I made a comment to a user asking about editors to use with perl. I was incensed by a remark he made about VIM not having the features he needed. Now, as I said in my comment, I've used VIM on a daily basis for over two years, and I'm still discovering new features — and I've used all of the features he was looking for.
This is where I discovered I'm a geek: my comment made it into the Daily Best for today, peaking around number 5. The fact that that made my day indicates to me that I must be a geek.
Oh — and VIM rules!
A new Cgiapp user reported they had stumbled across the project in php|architect! It's in the current, October 2004 issue, in the News section, prominently displayed in the upper right corner of the page. The announcement blurb is straight from my freshmeat project page for version 1.4. Cgiapp is carving a name for itself!
At work this week, I discovered a bug with how I was calling
set_error_handler() in Cgiapp's
run() method. Evidently passing a reference
in a PHP callback causes issues! So, I corrected that.
I also made a minor, one-character change to
query() to make it explicitly
return a reference to the
$_CGIAPP_REQUEST property array.
You can see full details at the Cgiapp download page.