There were lots of interesting discussion. Ed was attending for the first time, and is getting back into software engineering after some time in another career. He had some questions on what the different distros were and how they worked, and there were, of course, plenty of opinions. Susan had some updates on her research on the BF scheduler, the bleeding-edge Ubuntu releases, realtime kernels, and the Dragon Naturally Speaking application. I reviewed some of the upcoming meetings, and there was a lot of interest in the Seacoast LUG’s “Sugar on a Stick” presentation and the Cascading Stylesheet presentation at PySIG at the end of the month.

Mark McSweeney made the main presentation. Mark works in a small office with a few partners, and budgets are tight. A few years ago, they had deployed a Microsoft back end and discovered that there were no satisfactory solutions for spam filtering on the Exchange server they had as a mail server. Mark came up with a very effective and economical solution using PostFix, Amavisd-new, ClamAV, DCC, Razor, Pyzor and SpamAssassin. Mark’s slides can be found at http://wiki.gnhlug.org/twiki2/bin/view/Www/SpamFilter, including links to the solution he followed, an updated version of which can be found at http://www.freespamfilter.org/

Member Susan Cragin will be making the presentation at out March 1st meeting, on the Dragon Naturally Speaking 10 program running on WINE. Stay tuned for more details.

Thanks for Mark for his great presentation, to the NHTI and Library staff for the great facilities, to Dave Rose for bringing the projector, and to all for attending and participating!

Recently, I did some web development using XHTML 1.1 and CSS 2.1 and realized my 2004 version of the guide was getting out of date. I was pleased to see many of the pages had been updated to a 2009 version. After reviewing the many options, I chose to go all in and bought the Everything Book, a step up from my earlier version. This one includes cheatsheets for PHP, MySQL, JavaScript, DOM, HTML, CSS, HTML special characters, web colors and a great index. The reference not only includes broad coverage of each topic but many side notes and compatibility guides (for CSS, the IE-Netscape-Opera-FireFox-Safari compatibility color coding is tremendously useful!)

There are a number of bonus references available on the Visibone site at no cost. Check out the color lab, the color swatches for many of the common graphics programs, the online color codes reference, and excerpts from all of the various reference materials. In addition to reference book, Visibone offers posters, charts and mouse pads. The web site is worth a visit; it’s charmingly quirky, retro, opinionated and clearly individualistic.

<form...>
 <fieldset><legend>Ship To Address</legend>
 <br clear="all" />

 <label for="Building_or_Hotel">Name of Building/Hotel:</label>
 <input type="text" id="Building_or_Hotel" size="25" maxlength="50" />
 <br clear="all" />
...

</fieldset>

And the applicable CSS to make the form look pretty includes:

/* Styles for legends, the title atop the fieldset */
.dataform legend {
 background-color:rgb(137,111,43);
 color:white;
 font-weight:bold;
 margin-left: 0.5em;
 padding:0 1em 0 1em;
 }
/* Fieldset groups a set of inputs in a visual container */
.dataform fieldset {
 border-color:rgb(137,111,43);
 border-style:solid;
 border-width:1px ;
 background-color: rgb(93%,89%,78%);
 }

Results in a data entry form that looks like this:

Using CSS to add visual effects to the elements can yield some pretty sharp results while still having the speed and minimal bandwith of lightweight HTML and CSS text files. In addition, using the semantic fieldset and legend elements provides additional information about the form fields and their relationships in a machine-readable format, rather than just splattering a bunch of colored div boxes and styled text on the form. This is a great aid to tools used by the vision-impaired to provide more context to their users, and to search engines trying to derive the context of your forms.

Compatibility between different browsers is not too bad, though things could improve a bit. FireFox 3.5x, Chromium 4.0.285.0 (on Fedora, thanks, Spot!) and Safari 4.x appear dead-on the same (the screenshot above is FireFox 3.5.5), but there is an outlier and no one should be surprised to discover it’s Internet Explorer. Even in version 8, perhaps the most compatible browser Microsoft has shipped since the now-legendary IE5.5/Mac, things are just different. For some reason, the IE8 developers decided that the background color of the frameset should extend out over the border, to the edge of the legend, resulting in this less attractive appearance instead:

For IE7, the background color doesn’t appear at all, which might actually be better than to have it appear incorrectly. (Update: no, my bad, it just doesn’t like rgb values expressed as percentages.) So the appearance definitely won’t be as sharp for your IE users as for any other browser user. (Yet another reason to encourage your clients, friends, associates, neighbors and strangers on the street to avoid Internet Explorer. We can eliminate it in our lifetimes.) (UPDATE: there are, of course, work-arounds. Here are a few suggestions that might work for you from Sitepoint and CommunityMX and Matt Heerema.)

So, the next time you’re called upon to draw up an HTML data entry form, keep in mind that you can group your elements with fieldsets and use legends to add more contextual information to your design.

Title: Web Form Design: Filling in the Blanks

Author: Luke Wroblewski

ISBN: 978-1-933820-24-8

Published by Rosenfeld Media, 2008

In the nineties, I used to do a lot of presentations at conferences of computer programmers. I used to enjoy doing “concept” presentations; not so much focusing on “how-to” as to “why” and “what” to do. When I set the audience’s expectations correctly that they were going to be exposed to new concepts, I got a good reception. When I failed to make that point clear, I would inevitably get comments in my speaker’s review of “Great idea, but where’s the code?”

If you’re looking for the code, look elsewhere. This is not the book for you. If you are looking for the concepts, the big ideas behind user interaction on the web, how to get information from the browser, how to successfully convert the most browsers into customers, why different alignments might make a difference in accuracy, speed and satisfaction in filling out forms, this is the book for you. Luke Wroblewski is an experienced professional in the field of HCI (“Human-Computer Interaction”) or UX (“User Experience”) with some impressive industry experience. This book shows a thoughtful approach to “how do I get people to fill out a form, quickly, pleasantly, correctly?”

So as not to get lost in the weeds of HTML vs. XHTML and CSS 1, 2, 2.1 and 3 varying levels of compliance, Luke talks forms purely from the user standpoint: we see excellent cropped screenshots of forms, focused on the individual elements under discussion. (Rosenfeld Media has done an excellent job of producing a beautiful book: heft, thick pages, rich colors and fine printing.) The book is divided into three sections: Form Structure, Form Elements, and Form Interaction. Each section assumes some base knowledge, thankfully, and doesn’t start out trying to explain why you’d use a checkbox vs. a radio button. And he’s frank that there rarely is One True Answer (at one point, calling out, “All together now: ‘It depends’”) but talks about empirical testing of speed, correctness, and the results of studies monitoring user’s eye movements, pauses and mouse movements to determine which layouts work and which confuse.

If you spend a lot of time designing forms, the 215 pages of this book are well worth a read, and don’t be surprised if you find yourself going back to the book for a refresher.

Well, almost everything

Authors Rachel Andrew and Kevin Yank almost lost me at “hello” with their book titled, “Everything You Know About CSS is Wrong!” ISBN 978-0-9804552-2-9. I’ve never liked the “Dummy’s Guide” book for the same reason — I’m not a dummy — and I’m fairly confident (and hope my clients are well aware) that my CSS knowledge, while not encyclopedic, is better than average, and I’ve delivered some fairly good web solutions.

However, the first goal of selling a book is to get attention, and the title surely does that. And the opening line of Chapter 1, “The problem with CSS is that CSS is too hard.” Okay, I’m hooked, reel me in. (By the way, you can download the sample chapter 1 & 2 from the SitePoint web site; start at http://www.sitepoint.com/books/csswrong1/)

The reality is that the book shows one of the newest features, now available in all of the current brand of browsers, with the late release of Internet Explorer 8, of table layout options in the display attribute. After years of preaching that HTML table layout is less desirable than CSS layouts, this can be a hard sell, but Rachel and Kevin make a good case that a tabular layout of tables, rows and cells, but rendered from CSS and not HTML, is the best of both worlds. Chapters 3 & 4 show the equivalent layouts of many of the common design problems we run into (and perhaps fall back to using tables) and how they should work both with the new CSS 2.1-complaint browsers and how to fall back gracefully into a degraded but workable layout for earlier browsers. Chapter 5 gets into some very exciting layout possibilities that will be coming in the next couple of years as the CSS3 specification gels.

A big plus for the book is that it is brief: 111 pages you can get through in one or two sittings. The graphics clearly show the developing examples. The code extracts are clear (and all the sample code can be downloaded from the web sites). And the writing is clear and well-edited.

The book is well worth the $29.95 cover price, but keep an eye out around the SitePoint site (and their Twitter feed) to catch one of their frequent discounts.

The Thinkpads: T61 is the primary workstation, but was suffering from SMART-reported disk drive errors and I spent half the day following the awesome instructions for bad block HOWTO for smartmontools to rescue the disk. In the meantime, the T41 served as backup workstation letting me surf, check email, etc. In the afternoon, the A31p has a legacy WinXP/IE7 install I can VNC into to test a client project. Meanwhile, I was also working on the Lenovo ThinkPad W700 to finally fix the running-out-of-disk-space problem thanks to System Restore, now disabled (thanks, System Restore! pffft!). Why do people keep using Windows?

The three DVDs were to burn the System Restore disks once there was a machine worth restoring.

The two CDs were something old and something new. I tried a Knoppix 5.11 CD to boot the W700 so I could do a partimage copy of the Vista OS partition where we’ve installed custom software for client testing. For some reason, the older Knoppix couldn’t handle the hardware and would fail to start. I grabbed the latest Knoppix, 6.2 (make sure you grab the November 18th or later image) and – whoa! – what a difference. The 6.x series is based on different software and uses LXDE as the default desktop — very slick!

I used the iMac to burn the CD. The most recent software on the T61, Fedora 11, seems to be confused about whether I have a CD-R on the machine or not, and I haven’t been able to burn a disk for a while.

I used the second CD (in the T61) to burn an image onto a 1Gb USB Flash drive . The Knoppix 6.2 comes with a built-in option to create the USB with a mouse click or two. It boots fast, has 284 Mb left over to preserve settings, and boots in the W700.

Last night’s meeting was a vigorous and far-reaching discussion of MySQL, Oracle, the future of MySQL, Maria DB, OpenOffice.org automation using Python, OpenOffice.org automation using Visual FoxPro, Twisted, IE6, Zope, Plone, Django, MS SQL Server, pyodbc, SQLAlchemy, Cascading Style Sheets, IE6, FireFox and FireBug, User Agents, IE6, how not to insulate a bungalow roof, the (Python!) cssparse module (http://cthedot.de/cssutils/), Fortune’s selection of Steve Jobs as “CEO  of the Decade”, Lenovo netbooks and Ubuntu, the Millennium, why calendar years are one-based and not zero-based, distributed version control systems, master-slave and master-master replication using MySQL and Postgres, svn and git, and more! Whew! You should have been there!

Thanks to Bill for organizing the meeting, to all for attending and participating, and to the Amoskeag Business Incubator for providing the great facilities!

Stay tuned for an announcement of the December meeting, and hope everyone has a good Thanksgiving!

Brian was up first, and talked about rvm, the Ruby Version Manager (slides here: http://www.slideshare.net/bturnbull/rvm-nhruby-nov-2009). RVM doesn’t just allow you to have multiple versions of ruby on your machine; you’ve always been able to do that with side-by-side installs or linking or aliases or other kludges. RVM can be found at http://rvm.beginrescueend.com and can be installed as a Gem, from the Git repository or as a tarball. It’s not Ruby code, actually, but shell scripts, and it allows not just switching of the current active version of ruby, but much more interestingly, the entire set of gems associated with a version! Multiple gemsets can be assigned an alias that lets you group your gems by function or project or client. RVM also provides a facility to dump the list of gems from one version or gemset and load it into another, allowing you to test out the migration of an existing system to a new version with much less hassle. Finally, rvm also lets you invoke multiple versions serially, passing them a command like “rvm 1.86, 1.8.7, 1.9.1 rake spec” where rvm will run all your tests on multiple versions and report results, even in JSON if specified. This could be the core of a really simple Continuous Integration (CI) server in a single line of code!

Brian cautioned that rvm is early in its development cycle (current as of last night was 0.0.79) and changes are coming fast and furious, but it’s worth taking some time to keep track of this very interesting development!

Brian also mentioned that rvm has some extensions to work with Bundler. Nick mentioned that Bundler, a project from Yehuda Katz (core contributor to both Ruby and jQuery) , is slated to become the default means of managing gems in Rails 3.0. Here’s a good link to learn more: http://yehudakatz.com/2009/11/03/using-the-new-gem-bundler-today/

Nick Plante was up next, with his presentation on EC2 for Rails Development (http://zapnap.github.com/presentations/ec2-rubber/#0). Nick talked about the evolution of deployment of Rails (or Rack) solutions, moving from very painful to much easier with the introduction of Capistrano (http://capify.org/). He reviewed how simple it was to deploy an application with Capistrano, and then talked about the power of the Amazon Web Services (AWS) http://aws.amazon.com/ and Cloud Computing in general. Capistrano is focused on general deployment to a server, but the new add-on, Rubber (http://github.com/wr0ngway/rubber),  extends that to work with elastic computing resources like AWS’ Elastic Compute Cloud (EC2). We walked through the basics of the install-configure-deploy scenario and then “Insert Demo Here.” – he did it. Defying the Demo Gods (who had already tortured Brian a bit), Nick grabbed a copy of RedMine (http://www.redmine.org/) from GitHub, did the couple of minor configuration items needed to make it work locally (creating the session key, copying the sample database config to the live one, setting it to use SQLite3 locally and running db:migrate) and then set it up for deployment: first invoking Capistrano (doing some configuring) and then Vulcanizing (how else would you make rubber) the installation, then deploying it to a brand-new AWS instance. The steps flew by lightening-fast: the Amazon instance was up and running a minute after being invoked, then took a couple of minutes to install and configure Apache, Passenger, MySQL, the Redmine application, and sftp the development directory files to the instance. Within a few minutes, Nick could browse the machine and see the running application as well as ssh into the box and make any changes. There are innumerable options, of course, but they are all accessible by reviewing the configuration files and the source to tune it to your particular needs.

While we waited for the installation to complete, I brought up some of the customer’s concerns and practitioner’s insights from last Thursday’s MonadLUG meeting on Cloud Computing and we talked about how those applied. I expressed concerns about how persistent data storage worked, since the cloud images were considered ephemeral. Nick pointed out that the instance, while running could have a local database and it could be in a master-slave replication arrangement with either another instance (within the AWS, there aren’t bandwidth charges) or a hosted or local machine. Also mentioned was the Amazon Relation Database Service (RDS), which can provide a MySQL database instance with backups. So, there are lots of options.

Thanks to Brian and Nick for excellent presentations as well as organizing the meeting and bringing the pizza, to Tim and the folks at RMC Research for providing the excellent facilites and to all eleven who attended and participated in the meeting.

Remember, no Ruby meeting meeting in December. Stay tuned for a January announcement.

NHRuby is meeting for the last time this year on Monday the 16th at 7pm at RMC Research, 1000 Market Street, Building 2, Portsmouth, NH. Monday?! You say? Yes — due to scheduling conflicts, the regular meeting day for NHRuby is now to the third Monday of the month. Also, since next month’s meeting fell so close to the Christmas holiday, we decided to skip the December meetup and resume in January. So join us for the last meeting of 2009 for two presentations by NHRuby regulars, Nick Plante and Brian Turnbull.

Nick will present EC2, Amazon’s Elastic Compute Cloud. EC2 is a web service which allows customers to rent virtual server instances by the hour. The real power of EC2 is that it allows you to auto-scale your web applications on demand. Expecting heavy load today? Fire up another web app server or two, with no wait time for procurement, while taking advantage of Amazon’s robust infrastructure.

Making EC2 even more attractive are tools like Matt Conway’s Rubber, a set of extensions to Capistrano that allows you to script procurement and provisioning of server instances as well as deployment of your application itself. Need to set up a staging server? Use Rubber to deploy a disk image, install the necessary stack and utilities, and deploy the appropriate version of your application — in moments — all automatically. In this talk, Nick will quickly discuss the basics you need to know and then dive right into a realtime demo.

Brian will introduce RVM, the Ruby Version Manager. RVM is a command line tool which allows us to easily install, manage, and work with multiple Ruby environments and sets of gems. Topics to be covered include:

• Installation of RVM on Linux or OS X.
• Day to day use of RVM to switch between Ruby interpreters
• Managing sets of gems using Named Gem Sets

If you’ve ever been burned by differences between development and production, you should check out RVM — see how easy it is to take control of your Ruby environment.

So join us on Monday, 16 November at RMC Research for the last meeting of the year. Hope to see you there!