Vote Up
6
Votes
Vote Down
Mar 29, 2010 4:04 AM
6 comments

The assumption that a development team will be able to test all software used by their distro. Why let’s look at a real life example, Linux Mint:
I would like for you to look at the yellow box items, “Safe packages. Not tested but believed to be safe”

Levels of software safety

Take a guess which category a kernel update counts as. If you guessed level 5, pat yourself on the back for stating the obvious. This brings up a greater point, how a kernel, API and ABI can affect developers. Every time such an update occurs software has to be retested and reconfigured to work. Being that the Linux Mint development team cannot test all software before a kernel update is put forth; this creates a gap in trusted software. It is simply assumed everything missed will simply work until something breaks.

Inspired by the following quote:

I use the term “guardian angels” because most of us mere ordinary-desktop-user mortals lack the geek skills to evaluate the safety of software, much less the effect it might have on any particular operating system. Someone’s watching out for us! The Mint team covers my back, so I’m less likely to be posting one of those “b0rked by an update” requests for assistance. If you’re geeky enough to measure the risk of software and it’s effects on your o.s., this extra layer of safety may be bothersome. But it’s easy enough to add a repository and grab the older version of Synaptic with the gambler’s choice enabled. Be my guest.

Related Trademarks

#1 Posted by Delano on Mar 29, 2010 4:51 AM

In the Linux community, the community ARE the software testers. Duh. That’s why they say, “file a bug report”!

#2 Posted by Delano on Mar 29, 2010 4:55 AM

And seriously… we have yet ANOTHER religious analogy, and by members of the community themselves. Too disturbing.

#3 Posted by DrLoser on Mar 29, 2010 10:41 AM

I don’t see James Stewart taking the lead role in “It’s a wonderful Linux” any time soon…

Actually, when I think of “Guardian Angels,” I think of the guys who went round the New York subway system in the 1980s, beating up miscreants wherever they found them.

This sounds like a pretty good modus operandi for Linux Desktop testers…

#4 Posted by DrLoser on Mar 29, 2010 12:51 PM

One small other point. Apparently, level 1 is reserved for “Linux Mint or Romeo.”

I think I need the idea of distro packaging bludgeoned into my head, so’s I can understand the point of this.

#5 Posted by TheWHAMBurglar on Mar 29, 2010 5:01 PM

It is funny, important software such as Kerberos, DeviceKit-disks and Samba are not tested and are assumed to simply work. This just begs for a “What could possibly go wrong?”

#6 Posted by administrator on Mar 29, 2010 11:49 PM

As I’ve said before. The “usability testing” phase is just another name for “release” in FOSSland. Foist it on the users, let them break it and report.

It would be like a car company releasing a car and expecting the drivers to do their own crash safety evaluations.

You must be signed in to leave comments.