Michael Eriksson
A Swede in Germany
Home » Software development » Webdesign | About me Impressum Contact Sitemap

The manual checks for this website

Introduction/Page out-dated

The below discussion was correct until my then-computer broke in September 2009. After that I have neither re-installed all the tools mentioned, nor done any particular design work that needed extra testing. However, in the meantime Firefox 3.x and Opera 10.x have entered the mainstream, Chrome 5.0 is available under Linux, and non-exhaustive tests with these have taken place.

An interesting helper tool is http://browsershots.org/e, which makes screenshots using a wide variety of browsers and with some changeable options (including screen-size and JavaScript on/off). This cannot replace a manual check, but can help detect many visual problems.

What I check with

When developing this website, I use the following browsers and other tools to verify that pages look and function reasonably:

  1. Opera 9.x is my browser of choice, and what I naturally use to inspect the results.

  2. I regularly make test with Firefox 2.x; in particular, when working with CSS or when changing the HTML generation.

  3. CSS and generation changes are verified with Konqueror 3.5.7 (menus look a little weird, but are tolerable to me) and lynx 2.8.6 (pages degrade quite well, with exception for some ascii graphics and tables). Similarly, I occasionally turn of CSS support in Opera or Firefox to further check for degradability.

  4. Additional, but rare checks, are made with links, w3m, and (using a cell-phone emulatore) Opera mini.

  5. Similarly, occasional checks are made with an online simulator of various browserse.

  6. When making changes to colors, I use http://colorfilter.wickline.org/e to make sure that the page is at least readable with various color reductions.

(Also see my discussion of automatic checks).

What I do not check with

I do currently not test with Internet Explorer, Safari, and Chrome/Chromium: At the moment there is no good native way for me to bring them to action within Linux, and I do not have the time and resources to do tests under other platforms. Obviously, however, a corporate website should make very sure to make extensive tests with these browsers too.

Similar statements apply to other operating systems, various older and newer browser versions, etc.

Known problems

Apart from some minor quirks, I have two known problems:

  1. At the time of writing (2009-08-01) the only non-text contents are a few ascii graphics, which unfortunately do not degrade well in the form they are currently constructed. They will likely be replaced by real images and/or a different construction in the near future.


    Ironically, I originally went with ascii in the conviction that it was a safe solution that would work in basically every browser: Much less work needed than with images and better compatibility. Regrettably, this conviction turned out to be overly optimistic.

  2. Not all of the text based browsers handle tables well or at all; Opera mini (and presumably some other mobile browsers) has the same problem. I am unlikely to make changes here: Firstly, the text is still visible (even if misformatted); secondly, frankly, tables were introduced in 1995 (according to http://www.boutell.com/newfaq/history/fbrowser.htmle) and not supporting them in 2009 is border-line irresponsible.


    Note that the situation here is very different from e.g. Flash: There is no obvious reason why a user would deliberately turn tables off; implementing table support is comparatively easy; tables bring an actual benefit; both mobile and text devices have no reason not to support them in principle.

Design philosophy

Although I want this site to look good and be usable in as many browsers as possible, I am not striving for perfection—nor do I try to make every browser have the same display result: The textual contents are what matters to me, and as long as the readability and navigability of the site is not compromised, I am satisfied. (In particular, as I have reason to believe that the deviations and minor known problems are largely caused by incorrect implementations in individual browsers.)

Lynx and its relatives are given some special treatment to ensure that the pages degrade gracefully: They may not look good (but no worse than is unavoidable in a text based browser) and the usability is not compromised.