Archive for the ‘testing’ Category

Forza 5 asphalt bug

Wednesday, March 12th, 2014

Lately, I’ve been racking up the big bucks on Forza 5, by racing around the Indy loop. I set the drivatars to unbeatable ( and yet they all slam on the brakes before every banked turn ), turn the laps up to 20, and jump in the best car for the job (oddly enough, not the most expensive car available, but pretty close…any of the indy cars will do). It takes about 15 minutes for the race to run finish, and very rarely, I’ve seen a bug, where, and this is the part that the programmer and tester in me finds interesting, a piece of the track won’t get rendered, and displays as a flat black square. When it happens, I can see the black square approaching from the distance, and I’ve never seen one get re-rendered before I run over it, nor ever one fail prior-rendering once it’s made it into view.

It’s not a track it down kind of obsession, just a curiosity. Did they see it in pre-release testing, or is it a side effect of the real world that they weren’t able to recreate in-house? Is it caused by network congestion, local machine inadequacy, or some other unconsidered factor?

OpenSSH UTF-8/16 bug

Wednesday, September 18th, 2013

As I’m setting up a new music server, transferring files around with the scp that comes on OSX (Mountain Lion, I think?), and noticed that they have a tiny little bug in their handling of filenames with multi-byte characters. As files are transferred, a progress indicator is displayed, which leaves behind a line for every completed file. Usually these lines have several aligned columns, filename, percent complete, filesize, etc. But as a handful of classical pieces with french characters in the names go by, I see they are counting the multiple bytes, assuming they are all visible, so some of the columns were off for 5 files in a row.
As long as scp transferred the files intact, it’s a minor quirk/issue, but I do find it amusing, because I know I tested my old employers scp client with the same set of files, and it didn’t have this issue.

Appropriate Force

Tuesday, September 17th, 2013

Most of the time, I’m against forcing adults to do things.  But not always.

In designing a website, I think force should be applied frequently.  I’ll explain….

 

If you, for example, are building an e-commerce site, and are going to use a payment processor that requires shipping and billing information to match exactly, don’t put up two sets of fields and a warning that they must match.  Put up one set of fields, and one set of display-only fields that mimic whatever you typed into the first fields.  You avoid setting up your customers for failure, yet still salve the angst of the product manager who whines, “but people expect billing and shipping fields”.

At least that’s how I would do it.  But not a certain someone I visited today….and go figure, their system failed to work anyways, no matter what I tried.

Usually a Richard Branson company has better tech people than that, I’m surprisingly disappointed.

I was warned

Tuesday, August 13th, 2013

I was warned about getting back into PHP, but I didn’t listen, and now I have to solve silly problems created by the language trying to be ‘helpful’.

 

PHP doesn’t cast a string into an integer, unless it thinks it needs to, and vice versa.  If you are just working in PHP, that probably won’t bite you on the butt too often.  But if you need to integrate with another language, like say Java, because you are using PHP and selenium together in a BEHAT environment, you might get an error like,  Can’t convert java.lang.String to int

Which you have to solve by doing

function ITakeAnInt($mynumericstring)   // called with a paramter of  500;

{

  $mynumericint = $mynumericstring + 0;

  $something->javafunctionunderneath($mynumericint);

}

Seriously, if you can think of a less silly way to solve the lack of an explicit cast function in PHP, I’m all ears.

New job first days

Wednesday, August 7th, 2013

I’ve been at the new job for 3 days now, and am getting a feel for how things might end up going.

As birdie warned, and I am quickly remembering, PHP is a “a fractal of bad design”, quirky and cumbersome, but Behat really seems awesome in it’s potential.  I can’t help but keep thinking of how it could massively simplify my former co-workers’ job on the web application they are focusing on these days.  It seems plausible that I might actually manage to be writing new tests by the end of the week, at the new job.

The place is noisy, and the company small.  HR is still working their way through figuring out their hiring process, which leaves me uneasy about the potential for gaps in my medical coverage, but I’ve got till the end of the month.  People seem friendly enough, but very busy.

I remain ambivalent, but am at least feeling pretty sure I can accomplish what will be asked of me.

LG Doubleplay further suckage

Monday, July 22nd, 2013

So, I’ve made my peace with the craptacular way the LG Doubleplay is such a horribly QA’d phone that you have to hard-reboot it at least once a week, just to maintain it’s barely-functional level of service.  And by made peace, I mean, planning on getting a new phone as soon as possible.  But until then, I do the best I can with the phone I have.  Today was the day for doing the weekly reboot, only the phone locked up on the power-down confirmation message.  I could power a small city with how fast my eyes are rolling over this one.  Sorry if that intimidates you.

Google Play carries Julian Cope =)

Thursday, May 23rd, 2013

I’m not a big fan of the current UI and navigation flows, especially as relates to adding music to your library, but the fact that they have more Julian Cope than I do, has impressed me.  If they continue to have better coverage of my fave artists than Rhapsody, I could be in for a difficult decision in 30 days.  But of course, the second artist I search for, local musical genius James Coates, I find that, while they have his most recent album, which Rhapsody doesn’t, they don’t have his first album, which Rhapsody does….so it looks like the one-true-music-store has yet to be built; or rather to get the right legal agreements in place.

Really not a big fan of the way it kept stuttering and skipping, every time I searched, but maybe that’s just my local machine; I will try some other machines before I declare the play-programmers completely inept =p

Google+ photo failure

Friday, May 17th, 2013

With much fanfare, and free laptops, Google announced, among other things, significant ‘improvements’ to the wasteland that is Google+.  Of the stuff they touted, 99% of it sounded worthless and a complete waste of time, since they were features that are only useful if people actually started using Google+ exclusively, instead of Facebook.    The only new features that sounded worth checking out were the photo sharing related ones, so I fired up Chrome and logged into my + zone for the first time in months.

The first thing I see is that the UI design idiot who ruined Gmail, has had his hand in the UI for +, and made a mess of things there too.  Basic tasks that you can do with a single click on Facebook, take several to dozens of clicks.  But it sure looks ‘clean’ while doing nothing. eye-roll.

Anywho, the photo stuff…so first problem is, of course, they didn’t do any sort of QA on the uploader process, and it is very easy to crash, as a result.  As few as 240 pictures cause the uploader to crash.  I give the Chrome team props, that at least when the poorly written code from the G+ team kills the browser, it only kills the G+ tab, but still, it seems shamefully negligent to not have tested the uploader with the browser produced by your own company.

If you do manage to get some pics to load, the automatic animation builder works surprisingly well.  I’m still working on finding a stream of images that it will turn into a panorama, and will report those results when I have them.

Overall, it sounds like 2 good ideas, and 41 terrible ones, released before they were ready, by people who just don’t care about quality.  I doubt it will save plus.

Selenium stability

Friday, May 3rd, 2013

As we are approaching the first dot release of my first project tested with Selenium (and to a lesser extent, HtmlUnit), and I’m wondering what kind of projects it makes sense for.  It seems very unstable and brittle, requiring sometimes significant changes in test code, after a minor version update.