26 November, 2007

Django and Gantt Charts

It has now been almost a full week since I started the complete inkedmag.com site and infrastructure redesign in Python using the Django web publishing framework on FreeBSD, and I am happy to report that it is awesome. Mind you we're talking version .96 of the product, yet it truly is a dream with which to work.
I only recently started working with (and writing about) Cheetah, a wonderful python template engine, and had to very quickly learn yet another (Django's own template system). I must admit that Cheetah is easier to ready and learn quickly, but Django's system is considerably more agile in terms of conditionals and modifiers inside the template itself. There even happens to be a simple mechanism for cycling through a list continually changing on each iteration of the loop within which the cycle conditional resides. Simply put, it is wonderful for automatically changing the background colour of a row in a list.
For those unfamiliar with Django, it is simply one of the better web frameworks for content publishing on the web these days. While learning curve can be a little steep for some pieces of the framework, as a whole the speed at which once can produce working pages and applications is staggering. The ease and elegance of the system truly makes one enjoying creating new applications within the framework.
The first application I chose to migrate from native python into the framework was a simple store locator. The new version not only is considerably less lines of code, the database management was done for me at application creation/initialisation. I then simply exported the data from my existing application and imported it into the new table(s) Django created.
I could go on waxing poetic about every little bell and whistle, but I'd just be paraphrasing what many others have already pointed out online and otherwise. Don't think of it as Ruby on Rails because it isn't, though that isn't to be taken as an insult to Ruby. It is much more focused, cleaner and far simpler to setup and get running, including all of its own admin interfaces for the applications you create, as well as its own standalone development web server. Check it out, you won't be disappointed. This is going to save me a considerable amount of time.
Which brings me to my second point; Gantt charts. They are simply not something I find myself utilising on any regular basis, though I think that is going to change. I'm my own boss and have found that gantt charts produce the easiest visual way to show people the various pieces necessary for a project, when each portion can be expected to start and finish, all in parallel with the other projects for which I'm responsible (and/or coordinating).

I feel that the use of this tool more than others really gives a great method by which to see which projects will take the bulk of the time, and what projects overlap, etc. We have a system rewrite to produce and a whole server to replace, not to mention migrating certain custom software into the framework all before the new year. This is doable, but only because we've clearly set realistic (though tight nonetheless) goals and time frames. Consider using a gantt chart if you have more than one project or component of a project which needs to be done in a given time frame. Use one if you need to share with one or more people your schedule and need them to understand as quickly and clearly as possible that with which you are juggling or dealing. You find yourself quickly addicted to its usability.

12 November, 2007

Where Javascript Helps the User Experience.

     As is well known by a good deal of the regular readers of this blog, I have moved back into the world of being an independent Software Engineer, in an open ended contract with Pinchazo Publishing Group, Inc.  Their best known publications are Nylon (featured recently in the newest iPhone commercials from Apple), and the recently re-launched Inked, a tattoo-culture centric magazine, both of which are distributed globally.  

I bring up all of these specifics because it marks a decidedly big shift in my own coding career.  I have traditionally worked on back-end and middle-ware systems, making incompatible systems play nicely together, hardly have I ever had to deal with front ends and end user interactivity.  Sure, I did the web page for Thelesis, a non-profit group, including the framework and almost all graphics, and continue to maintain that site to this very day.  As a whole though, I never felt a desire to deal with the front end, I like the logic behind the interface point of view.  Well, now I'm in a situation where I'm needed to make tools with which end users will interact primarily.  Odd change eh?

It was at my previous employer, Blue Gravity Communications, a wonderful FreeBSD centric (with some Linux) hosting company that I found myself needing to really start to learn Javascript in order to convenience the end users in the selection processes.  It was here that I started to learn more about a language with which I never thought I would have a need.   I couple this to mention from a good friend about a post from a wonderful developer (my aforementioned friend's previous co-worker), regarding how wonderful javascript can be in one's toolbox.

With all of this in mind, I needed to jump into the world of user friendly interfaces.  I know from my own experiences perusing the web that I know what a non-intrusive interface is like, but it really isn't best to ask developers what a good interface is all about.  By nature, we are far simpler in our needs and all too willing to overlook certain practises that we don't see as a problem.  Keep in mind, many developers, myself included, still prefer command line interfaces because of how much quicker they generally are.  

I've already called upon javascript for certain pre-submit form checking, which is ultimately a convenience to the end user because it saves them having to reload the page, or worse off, play hit and miss with multiple loops of the process of submitting and seeing what was wrong with their form submission.  This is a very unfriendly approach in 2007 which is sadly still utilised by many large web based corporations.  

This time was different as I was coding the first version (what I would normally tag as a beta) of Inked magazine's online tattoo gallery.  The concept is simple, allow users (and store owners) to upload tattoo photos for general viewing on the website, and if a specific photo is from a tattoo shop/parlour in our tattoo shop database (covering 4 out of every 5 shops in the United States), make a link to that shop so that browsers of the gallery can associate certain quality work with a given producer of body art.  Very simple, great use for your standard LAMP (in this case BAMP [BSD, Apache, MySQL and Python]) configuration. 

However, end users could care less about the underlying technology, they care about ease of use.   It was with this mentality in mind that I approached the gallery's first incarnation.   Limit the amount of non-photo graphics (for speed), limit the amount of time a page actually needs to be refreshed and/or requested, and make the controls relevant and simple to understand.

This was (I believe successfully) achieved by use of a strong reliable template engine for the purpose of controlling what user control elements were presented for navigation on any given screen.  Ultimately, if a person is browsing paginated libraries of content, we only wish to have he navigation controls relevant to where said individual is in their browsing activities, visible.  Meaning that if a person is on the first page of a five page gallery, don't render the button that links to the first page, and don't render the button which links to the previous page (as it is non-existant).  Likewise, we don't want have buttons for the "next" page, or the "final page" when we're actually on it.  This may seem logical, which I'd like to think it is, yet so many seem to overlook these kinds of details.  These are details which can cause frustration from users who unintentionally click on a button which goes to the same page they are already browsing, or in the case of a "ghosted" button, make them wonder why it isn't working at all.  Only present that which is needed, and nothing else if at all possible.

More importantly, and even less obstructive, javascript for auto-zooming the photo gallery images themselves without having to pop-up a window, or even worse, replace the current viewing page with a simple image link or dedicated page with headers and footers in addition to the image.  These elements are time killers, and javascript is one wonderful way in which to resolve the problem.  Not only does this kind of visual add an interactive feel to the page(s), far more similar to the way a user would experience their own operating system (especially these days with candy like OSes), but it means they aren't hindered by unnecessary delays and can focus clearly on that for which they came to the page in the first place.  To view photos of tattoos that interest them, or share theirs with the world.

07 November, 2007

Cheetah, Python's Powerful Template Engine

About six months ago I wrote an entry about using the Template Toolkit for perl, and how I found that it was almost as if giving perl a little taste of Python. Now, fast forward to present time and I find myself as my own boss once again and in a dedicated open-ended contract with Pinchazo Publishing Group for Nylon Magazine and more recently, Inked Magazine. This opportunity has also proved to be beneficial for me in that I get to choose the technologies with which to arm these businesses moving forward for their presence on the internet.

This brings me to some realisations to which i came today. Python's template engine "Cheetah" is considerably better than aforementioned Template Toolkit for perl. I'm currently writing a new online gallery application using Python, MySQL, Javascript, CSS and of course HTML on a BSD server running Apache 2.2. Today was the first actual coding day for implementing my design, and while there were certain changes of some underlying routines, I have to say that it is moving along much quicker and smoother than alloted/anticipated. I attribute this heavily to the ease of use found within the Cheetah library.

Some template engines add a quasi familiar set of language constructs which make using such a system doable but with that kludgey feeling. That is not that case with Cheetah and in true Python fashion, it integrates using constructs that closely parallel the standard Python syntax, as well as offering several additional alternatives to help adapt in various situations and code bases.

The beauty of using template system (as has been said before) is that you add an additional layer of separation of code from display to the point that in team/diverse environments, the coders and artists don't interfere with one another. A simple protocol of self-discipline for each individual to stick to their roles ensures that both content and display functionalities can be developed, and changed simultaneously without concern over coordinating the end result. The busier the schedule, the crazier the deadline, the quicker (and with a much higher level of confidence and lower level of stress) that a project can be implemented/modified/redesigned.

Once the gallery is fully operational, I will be updating this post to add a working link to the site. It should only be a few days from the committal of this blog entry, so keep up to date by subscribing via the codedevl rss feed (courtesy of atom).