Fwd: Re: Helping to improve advertising of test days and other things

Adam Williamson awilliam at redhat.com
Thu Aug 23 20:07:08 UTC 2012


On 2012-08-23 0:31, Arnav Kalra wrote:
> Would it be possible for me to edit someone elses results in the 
> wiki?
>  If yes then we need to have a system for authentication. The reason
> why smolt came to my mind was because they have a system for
> authentication.
>  When I send my profile using smolt it makes a blank page on the
> website which contains the template for results. If I want to edit it
> I use smolt on my system to generate a key which I input into that
> page. This allows me to edit it and makes sure that only the person
> with that installation can edit it.
>  We can do something similar for testing. For each boot a page can be
> created automatically  and the person who is testing is allowed to
> edit it. To avoid duplicate entries we can allow a person to link all
> those authentication keys or pages with his fedora account.
>  Is this viable?

This is all perfectly true in theory, but to my knowledge has never 
turned out to be a problem in practice. I don't think we've ever had a 
case where anyone edited anyone else's results on purpose. It has 
happened accidentally on occasion, but it's still not a huge problem. 
For test days it's just not terribly important: we don't need test day 
results pages to be 100% accurate, they really just give us a broad 
impression of what's going on, and 99% accuracy is fine for that. Any 
actual failures are meant to be backed by bug reports, which can't be 
'overwritten'.

For release validation the accuracy of the results tables is much more 
important, but for exactly that reason, any errors get caught. Several 
of us regularly look at the results page for each release validation 
event and double-check any listed failures; there's intensive manual 
review of the pages and we do catch any problems. Remember, the wiki has 
a big strength in revision control, to offset the weakness in access 
control. We can very easily see the history of all changes to the 
results pages and catch any errors, and easily revert to earlier states 
if necessary.

I'm not actually convinced that implementing access control would gain 
us anything at all practical for test days or release validation. Like I 
said, in theory it's utterly crazy that anyone at all can come and 
change anything at all on these pages. In practice, it turns out not to 
be a problem.
-- 
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora
http://www.happyassassin.net


More information about the test mailing list