Interactive | Fiona MacNeill | Macknowlogist

Striving for accessibility - Eval pt 1

Thu, 19 Jan 2017 00:00:00 +0000

Part 1 of my final evaluation. This entry outlines key decisions that I made regarding accessibility towards the end of the development period. I am writing and publishing it retroactively as it as represents a very large amount of thinking and work, which I kept notes about as I went along.

Based on observation

Based on looking at a large number of NHS and governmental sites, referenced in former posts I realise that accessibility is crucial, perhaps even more than your average portfolio site. For example, looking at NHS England (2017) site I see that they offer a text-to-speech service. Based on experience in my day-job working with software vendors, some of these solutions can be quite expensive. It was however important to me to make the site as accessible as possible and that has meant some sacrifices along the way.

Keystone texts

While engaging in this in-depth investigation into the accessibility, I found myself referring to the following keystone texts.

- UX accessibility specialists an amazing resource
- a frequent go-to of mine

I feel like I have only scratched the surface with the WAI-ARIA roles, but based on testing I feel that the site is accessible as I could get it within the allotted time. In the end I found a blog post on the subject by a rather brilliant developer, Aaron Krauss (2016) who did a great job of how to practically use certain ARIA labels and roles. What I did find is that there is quite a lot of redundancy in the code, but you need to have it there as the screenreader doesn’t necessarily pick up the semantics in the page construction as you might intend. Here is a video of my screenreader testing in action with tab key only - tabindex (Mozilla Developer Network, 2017)

What the font?

I conducted some reading (Nielsen, 2002; Martin, 2009; Franz, 2014; ) around legible fonts and font sizes and optimal line height. I have tried my best to work with the best practices, although that does get tricky with the mobile screen size. However, the attention that I have paid to optimising the CSS stylesheet for plain readers should help those who use accessibility features on mobile devices. Added a fix to the superscript I found a very helpful snippet for fixing line height for superscript (used for references) and subscript. Thank you to the -ever helpful- CSS Tricks site (Coyier, 2009a). Contrast As much as I loved the white headers along the timeline the contrast was not sufficient. The colour scheme is themed around blues and purples and unfortunately white on blue doesn’t comply with WCAG 2.0 unless the blue shade is very dark or the white font is very large (Snook, 2015; WAVE, n.d.).

The accordion had to go

Upon testing the tab-key only navigation I realised that there was no accessible way (that I could muster anyway) to navigate the accordion (Mary Lou, 2012) and for that reason it had to go. I added two simple tables, which I was trying to avoid, but as they area features comparisons this use us permissible. I was not using the tables for layout, but for the display of information in a standardised way. After the fact I found this accordion example at codepen.io which might have been better, but not without a lot of work:

Sprite fun

I got my sprite on and added sprites for the device images. The device images were drawings that I created in Adobe Illustrator based on photographs. All the photos were apart from the image of jeans, which I have listed below. This was added as an image through the CSS and aria-labelled for descriptive purposes.

The complete sprite for the devices in all of its glory. Photograph used as the basis for illustration of fasten/clip wearable - by is licensed under CC0 images sources from Pixabay:

Fixing the tab svg

I replaced the svg tab to be a proper thing rather than the Frankenstein’s monster-like CSS creation . Once again I referred to the excellent CSS-Tricks (Coyier, 2013) and Mozilla Developer Network resources (2016c), as well as the Implementing Responsive Design book from the reading list (Kadlec, 2012).

Asking myself questions

I tried my best to standardise all the units in the CSS although if I were to do it all again with hindsight I think that I might design my grid to work with ems or rems (once they are widely compatible or older browsers fall out of use). I’m sure that I have missed things in spite of several thorough read-throughs. It all validated okay. CSS code block structure: e.g. div class or id - this resembles the layout of a div .div { display: block; position: relative; width: 90% height: 60% color: #000; margin: 2% auto 2% auto padding: 2% 4%; (shortenings used where possible) } e.g. typography class or id - this resembles the layout h2 { font-family: “Helvetica Neue”, Helvetica, Arial, sans-serif; text-align: center; font-size: 1.5em; line-height: 1.25; font-weight: 300; } In the HTML where white-space pre-line formatting was needed for the Goals and the references sections, I included comments to help indicate why the formatting was different in the markup. Also regarding the CSS as I was going through it, to weed it a bit, I asked myself a series of questions to help determine which units to use and such. I have included these below.

Is it text? = ems
If it moves it is em or %. If it doesn’t or cannot move it is pixels
Is it in a column in the grid - then margins and padding are percentages (unless there is a very good reason for them not to be)
Is it in the grid - then margins and padding are percentages
Is it used in the html? No, then get rid of it Could it be consolidated in any way? Does it share the same values as another class and/or id?I would have liked to do more on this, but I ran out of time. Did my best to do it as I went along.
Does it need to be a negative value (particularly in the media queries)? Then it is in ems.

Rationale:

When it is something that needs to have set dimensions such as a button, I am using pixels.
For text I am using ems to ensure compatibility with older browsers
For table related items and blocks which reside in divs I have used percentages.
Used hexadecimal shortenings when it made sense to do so and it did not change the hue of the colour (three digits rather than the standard 6). I prefer digits rather than word descriptions, due to colour precision.

Media queries

Did a lot of work on these, again I found the Implementing Responsive Design (Kadlec, 2012) and the Introducing HTML 5 (Lawson & Sharp, 2011) books particularly helpful. I also added a arrow keys image and tooltip to help explain the navigation of the timeline. You will notice for the tabbed navigation of the site in the screen reader demo that the timeline is skipped over initially and I added to the JavaScript in order to allow the tab key for navigation in addition to the arrow keys. The arrow keys allow for greater freedom though, as once you have started tabbing through the carousel/timeline, you cannot get out of it with another tab press. Thus leaving it to the end and this is something for me to raise with the original developers to see if they can address that in future versions of the widget.

References

Coyier, C. (2009a, August 10). Prevent Superscripts and subscripts from affecting line-height

\[Blog post\]

. Retrieved from

Coyier, C. (2009b, October 24). CSS Sprites: What they are, why they’re cool, and how to use them

\[Blog post\]

. Retrieved from

Coyier, C. (2013, December 4). SVG tabs (using an SVG shape as template)

\[Blog post\]

. Retrieved from

Donut. (2011). Re: What are the most common font-sizes for H1-H6 tags. StackOverflow website. Retrieved 19 January 2017, from

Franz, L. (2014, September 29). Size matters: Balancing line length and font size in responsive web design – smashing magazine

\[Blog post\]

. Retrieved from

Kadlec, T. (2012). Responsive Media. In Implementing responsive design: Building sites for an anywhere, everywhere web (pp. 95–127). Berkeley, CA: New Riders Publishing.

Krauss, A. (2016, September 8). ARIA roles and attributes: How to actually use them | Aaron Krauss Retrieved from

Lawson, B. D., & Sharp, R. (2011). Introducing HTML5 (2nd edition) (2nd ed.). Berkeley, CA: New Riders Publishing.

Martin, M. (2009, August 20). Typographic design patterns and best practices – smashing magazine

\[Blog post\]

. Retrieved from

Mary Lou. (2012, February 21). Accordion with CSS3

\[Blog post\]

. Retrieved from

Mozilla Developer Network. (2015, July 23). Using the aria-describedby attribute. Retrieved January 19, 2017, from Mozilla Developer Network website:

Mozilla Developer Network. (2016a, December 15). WAI-ARIA basics. Retrieved January 19, 2017, from Mozilla Developer Network website:

Mozilla Developer Network. (2016b, November 17). CSS values and units. Retrieved January 19, 2017, from Mozilla Developer Network website:

Mozilla Developer Network. (2016c, December 3). Background-position. Retrieved January 19, 2017, from Mozilla Developer Network website:

Mozilla Developer Network. (2017, January 16). Tabindex. Retrieved January 19, 2017, from Mozilla Developer Network website:

NHS England. (2017). About NHS England. Retrieved January 23, 2017, from NHS England website:

Nielsen, J. (1996, October 1). Accessible design for users with disabilities

\[Blog post\]

. Retrieved from

Nielsen, J. (2002). Let users control font size

\[Blog post\]

. Retrieved from

Snook. (2015). Colour Contrast Check

\[Computer software\]

. Retrieved 19 January 2017, from

Stanford University Online Accessibility Program. (2016, August 1). Screen reader testing. Retrieved January 19, 2017, from Stanford University website:

W3C. (2016, October 27). Accessible Rich Internet Applications (WAI-ARIA) 1.1 W3C Candidate Recommendation 27 October 2016. Retrieved January 23, 2017, from W3C website,

W3C. The Roles Model. Retrieved January 19, 2017a, from WAI-ARIA website:

W3C. WAI-ARIA role definition model - image. Retrieved January 19, 2017b, from WAI-ARIA website:

W3Schools. CSS Tooltip. Retrieved January 19, 2017, from w3schools.com website:

Watson, L. (2014, August 4). Using the tabindex attribute

\[Blog post\]

. Retrieved from

WAVE web accessibility tool.

\[Computer software\]

. Retrieved January 19, 2017, from

The weekend of doom

Mon, 16 Jan 2017 00:00:00 +0000

In my last entry I talked about the suggestions that Theo provided from his perspective as a nurse and researcher. They were very helpful, but lead to some rather major layout changes. In particular, the to do: “I will need more space in the carousel/timeline”. This led to what I nicknamed in my GitHub posts as the Weekend of Hell, I am softening this within the context of this journal entry to be the Weekend of Doom.

. An image showing how the site looked prior to the changes to the carousel.

Altering the Flickity Carousel/Timeline

So changing the height of the Flickity carousel used for the patient timeline meant that the “dot” buttons had to be moved upwards otherwise they would be off the bottom of the visible area on most standard screen sizes. For reference, I have included a long screenshot that I made of the site prior to this work, on the right. So I decided that the dots should reside on-top of the carousel and then they would also be spatially connected to the navigation of the timeline. This is not really how Flickity carousels are really designed to exist (metafizzy, 2016), but it is closer to the original inspiration for my site the DuckDuckGo about page (DuckDuckGo, 2014). The main issue that I ran into was that the carousel needed to resize itself dynamically and I had media queries which hid the dots on mobile devices (as they were no longer very useful as buttons at that size). However, as the dots themselves are rendered automatically by the JavaScript, they were rendering much more quickly than the resizing. So I upon sizing-up a responsive browser window the dots and timeline appeared on top of everything and looked awful. I tried everything I could think of to fix this issue. I attempted to delay the dots in the JavaScript. I also tried writing a script which watched for the window size. I tried positioning the dots so that they moved upwards. I also tried creating condition code which watched for various conditions to be met prior to triggering the dots. Things I wrote worked to an extent, but not sufficiently or reliably enough to depend on them. In the end I had to temporarily hide the dots by setting them and the timeline to ‘display: none’. I also had to hide the spot where they generated by moving the about section upwards. Essentially the solution’s analogue equivalent is that the rabbit is being hidden in a compartment of the hat the whole time (oh wait…is that how they do it?).

Performance Issues

I am actually grateful that all of this happened as once I had cracked the rendering issue I realised that there were some major performance issues with the carousel/timeline. It took a bit of time to get to the bottom of it, but in the end I realised that it was the long svg that I had included to indicate the patient’s anxiety levels. Through testing in multiple browsers, on Mac, PC and mobile I realised that the image also wasn’t loading anywhere other than in Chrome for Mac. Everywhere it was just gumming up the works. I reassessed the need for the anxiety level gauge/image and decided that it was important to keep it. So through trial and error I ended up loading with an image tag in the html as a very, very long svg. This is the only svg in the site that I could not optimise as it resulted in optimisation errors because it was too long dimensions-wise. Also, I realised that I did not really want to optimise it as I would result in losing the drop shadow detail which I had added to a new version of the design.

The new design

The performance issues highlighted that the design of the patient anxiety-level link was in fact creating an optical illusion and making the carousel look guddery/jerky even once the problems with it were fixed. The new design not only looked more streamlined, but it moved more smoothly simply by virtue of its aesthetics.

Version 1 - This screenshot deliberately shows the progress between two slides. This version of the anxiety level image had a visual effect akin to passing a row of trees with bright light behind them. It became very jarring when viewed at-speed

Version 2: I created a long svg in Adobe Illustrator with a greater level of logic. Attempting to map the story thematically to a graphical interpretation of the patient stress levels. This link is on the one-hand more graph-like but also more oraganic as I made it from scratch by drawing and manipulating a continuous line…Can you tell I went to art school?

Space economy

This entry represents the end of a period where I added a lot of content to the site. This was a good thing, but it also meant that there were areas that felt overly heavy like the devices area. I decided to try adding a collapsible accordion as a container for the information, using a tutorial from Codrops (Mary Lou, 2012). The tutorial was bit old but I liked how it used CSS and I felt that I could build on it. The accordion looked quite nice too, I’m sure that you can sense a ‘but…’ coming on here? Well I will explain it in the next journal entry. I have however included a screenshot the accordion so that it can live on it perpetuity.

. This is where I got to with them. Partially styled, but a nice efficient use of space. I would have added svg “+“s on the right to indicate their interactivity.

The contact form actually works now

I wrote a dev post about the contact form, but yes the contact form is now working as it should. I just need to enhance the warning/alert aspects and add aria labels etc. And so does the back-to-top the button as in it only appears when one is sufficiently far from the top of the page for it to be useful. This took a couple of iterations of code to get right.

Finally…

I had come up with a CSS style for reference sections and I decided to add it to this journal as well as part of my custom child theme. I am going to try (time allowing) to add it to my all my posts retroactively, but it is a big job and I have surprised myself at the sheer number of posts that I have amassed here in the past few months.

References

Coyier, C. (2013, March 5). Using SVG

\[Blog post\]

. Retrieved from

DuckDuckGo. (2014, February 20). About DuckDuckGo. Retrieved January 15, 2017, from DuckDuckGo.com website:

Mary Lou. (2012, February 21). Accordion with CSS3

\[Blog post\]

. Retrieved from

metafizzy. (2016). Options. Retrieved January 15, 2017, from metafizzy website:

Sowmya. (2013). Re: Nesting an svg inside a div

\[Online forum comment\]

. Retrieved from StackOverflow website:

Charty McCharterson

Sat, 07 Jan 2017 00:00:00 +0000

I knew that all of that work that I did on polls would come in handy at some point. The time arrived, as displaying data was something that seemed important from the beginning, to support the goals for the project. I wanted to support the information about ‘UK smart device ownership’ with some data. Having spent quite a bit of time and trying things out, I felt sure that the the Google Charts api (n.d.) would give me the level of interactivity and customisation required.

Something I changed

Like the tools I had used already (see earlier entries including Flickity by metafizzy (2016), smooth-scroll by cfernandi (2016)) the instructions always say to call the JavaScript package in the header. This is something that my instructor, Marcus, had pointed out as bad practice. Therefore, I created a new eventlistener in my JavaScript file in order to load google’s “loader.js” script. This allowed me to place the ’ tag at the bottom of my index.html file. However, unlike the other widgets and code snippets that I have used, I cannot compile Google’s JavaScript as part of my own script (Google, 2016a), it has to be loaded from Google servers (licensed under Apache license version 2.0 (2004) ). For this reason I reused the loading GIF from the contact form, to show while the chart loads. The data itself is from Offcom’s report entitled, The Communications Market, 5 Internet and Online Content (2016a, p. 191). What is great about their site is that they make the raw data available in Microsoft Excel format, so I was able to construct the chart based on a subset of data (Ofcom, 2016b). In order to show the age versus ownership I selected a stacked barchart as a space efficient way to show the data.

Prepping the chart for print

I have still to create my print stylesheet, which is something that I will do towards the end of the project. In the meantime it made sense to make sure that the chart was prepped for printing to png. I am not going to make a link to the png available as that function will run in browsers which support the HTML5 element (Google, 2016c). However, what customising this script did do, was create a static view of the chart showing it in suitable format for printing.

Tab clear

I decided that I needed to highlight the User Experience section of the portfolio page design, as it was all very wordy. There is still a lot to consider in this area, but I decided to add a tab image on the top-left of the UX factors (formerly scenarios) section. I started by creating an svg, partially inspired by a post on CSS-Tricks (Coyier, 2013). However, I really wasn’t happy with it in terms of formatting and scaling. For now I have replaced it with a tab which I created using pure CSS. The tab is actually made of two divs, one for the the main tab shape (rounded rectangle) and one for the slant leading up to it using the skewX transform option. However, I will have to replace this as it is only compatible with IE11 or later (Can I use, 2016). It is a shame, but it was good to learn about the transform option and again I used the ::before pseudo element to add text content on the div.

The first attempt at the tab using an .svg

Second attempt at a tab using two CSS divs - one skewed 40deg - sadly it will need to be replaced due to the need for background compatibility. Ultimately, I will need to go back to an svg solution, but I will draw it to look like what I came up with using the CSS.

References

Apache Software Foundation. (2004). Apache license version 2.0. Retrieved January 7, 2017, from Apache Software Foundation website,

Can I use… (2016, October 23). Retrieved January 7, 2017, from

Coyier, C. (2013, December 4). SVG tabs (using an SVG shape as template)

\[Blog post\]

. Retrieved from

cferdinandi. (2016, December 11). Cferdinandi/smooth-scroll. Retrieved December 11, 2016, from

Google. (n.d.). Charts | Google developers. Retrieved January 7, 2017, from Google Charts,

Google. (2016a, September 12). Load the libraries. Retrieved January 7, 2017, from Google Charts,

Google. (2016b, September 12). Bar charts. Retrieved January 7, 2017, from Google Charts,

Google. (2016c, October 12). Printing PNG charts. Retrieved January 7, 2017, from Google Charts,

metafizzy. (2016). Flickity. Retrieved January 15, 2017, from

Ofcom. (2016a). The Communications Market 2016: 5 Internet and online content. Retrieved from

Ofcom. (2016b, August 4). Data downloads. Retrieved January 7, 2017, from Ofcom.org.uk,