f-e-k improvement discussion continued

[Adam Williamson]

On Fri, 2010-07-16 at 23:01 -0400, Bob Lightfoot wrote:

>  BobLfoot's next comment: I just ran yum update
> --enable-repo=updates-testing --skip-broken -y and when I went to run
> f-e-k I was presented with three packages to provide feedback on.  1)
> ppp 2) xorg-x11-xinit 3) xsane-common and xsane-gimp. 1) I've never
> used ppp networking so this would be a package I'd of karma 0 and "not
> tested" comment prior to Adam's request that we not do so.  The 

Actually, ppp is a critical path package, so going by what I wrote, it
would be okay to leave a 0 karma 'didn't break critical path but not
directly tested'. (Although actually, for ppp, we really need feedback
from people who actually use PPP connections).

> suggestion to "just remove the package" is unacceptable in this case
> also since yum erase ppp wipes out Network Manager. So not testing
> would be a nice choice here for me rather than having to skip it every
> time.  a karma of I or 7 which didn't factor into the total karma math
> but caused the system not to represent works great. 2) I run x on my
> VM but don't know if I have an Xorg file or not and wouldn't have a
> clue how to test for a missing bang to shebang unless the developer
> choose to publish a test protocol.  

Again, a critical path package, so it's fine to leave a comment here.
And yeah, that's a terrible update description, just awful. (Sorry,
mcepl :>)

A shebang is one of those bits at the start of a script that looks like
this:

#!/bin/bash

which tells the system what shell the script is supposed to be run with.
The shebang is the #! part. The # is the 'she', the ! is the 'bang'. So
this means some script in the package had been accidentally left
reading:

#/bin/bash

and the update fixes that. It doesn't explain *which* file has the fix,
though, or what the actual impact of it being broken was. Or provide a
bug number.

> So not testing would be a nice choice here also, but I guess I'll just
> have the annoyance of skipping this going forward. 3) Xsane will get
> tested before I provide feedback,  I have an HP flatbed connected and
> will exercise xsane. Small sample but must pass over 2/3 to get to the
> one that counts. Does anyone else encounter or follow my point? 

In general, indeed, you've explained pretty well why sometimes you'd
have a package installed but not be able to directly test it.
-- 
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Fedora Talk: adamwill AT fedoraproject DOT org
http://www.happyassassin.net