Kevin Daly

Yes folks, I'm testing blogging tools again. This time it's Word 2010…Yeay.

A little while ago I finally released (as noted in an earlier post) the Windows Phone version of Diarist, my mobile blogging application.

Also as previously noted I’d decided at the last minute to release it as a free application, partly because looking at things in the cold light of day caused me to realise that the market for such an app was never going to be large enough to bring in enough to justify the effort of filling in the forms for provisional tax (if you’re unfamiliar with New Zealand tax law, count yourself lucky), and also there are issues such as the  somewhat unsatisfactory  TextBox scrolling experience on WP7 which left me uncomfortable with the idea of charging money for a text-heavy application – and to top it off the Lodsys patent trolling situation currently afflictiong iOS developers (and now Android devs) left me unsure about the viability of Trial Mode.

This turned out to be a lucky decision, because I allowed a howler of a bug to slip into the initial release, and also through sheer carelessness provided a description that meant users would fail to be able to find how to use one of the features that really takes advantage of the OS, and would instead simply assume that it doesn’t work. Mea maxima culpa.

A day or so after after Diarist went live I received some feedback (the only direct feedback I’ve had so far) from a user who noted that despite the app description and the relevant blog post saying that there was integration with the Pictures hub via the “extras” menu, no such integration existed from that location, and also that line feeds and double line feeds failed to be converted to HTML line break and paragraph tags.

Gulp.

The first of these issues was sheer carelessness on my part: I’d relied on my memory for which menu option provides the “upload to Diarist” option without checking, which is completely unforgivable (it’s actually “share…”, not “extras”).

The second issue was a bug which found its way into the release because of a failure to do regression testing.
I had originally been specifically testing for the newline (or rather, and this turns out to be significant, return) characters used in WP7 TextBox controls, and that’s the code which I had tested for conversion to HTML markup (<br /> for a single line break and <p /> for two). Shortly before release I decided to replace this code with Environment.NewLine (a system constant which theoretically represents the newline character in any given environment) as the correct way to use an appropriate abstraction. This turned out to be a mistake, because once I’d been informed that returns were not being converted I found that a Return in WP7 (at least from the SIP) generates a \r – in other words a carriage return, whereas Environment.NewLine has the Windows newline value of \r\n (carriage return + linefeed). So I’d introduced an untested regression by making an assumption about the value of Environment.NewLine on WP7. As all developers know, “assume” is as bad as the “S”-words (“Should” and “Surely”).

So the lesson was one I should not have had to learn, having known it for many years: when you change anything, be sure to do regression testing.

PS. The corrected version is now in the Marketplace – which sadly will be of no interest to all the people who encountered my original shoddy release, justifiably concluded “This is a load of crap” and uninstalled.
So I’m treating that one as a learning experience.