Package Review Stats for the week ending January 18th, 2009

[Thomas Moschny]

2009/1/29 Nicolas Mailhot <nicolas.mailhot at laposte.net>:
>>> I actually knew the exact file source Sven would use because he had
>>> packaged a related font from the same upstream two days before.
>>
>> No one is really doubting that. But, it's all about *documenting* what
>> you did (and what you didn't) in the review. That's the most important
>> reason for the existence of a a review ticket at all.
>
> The most important reason for the existence of a a review ticket is to
> make sure someone in a trusted group did review a package.

Even trusted people can make mistakes, oversee something, or
guidelines change over time, etc. Having checklists is generally a
good tool when a task is complicated and/or consists of a series of
subtasks. I'm not ashamed using checklists in such cases, but of
course I understand not everyone else uses them.

> In case of doubt the information in the review ticket is just as
> likely to be flawed as the package itself, and in the end the package
> we ship is the only thing that really matters.

Agreed. But the fact that there's no enforced connection between what
was reviewed and what is imported later on, is basically a flaw in the
tools we use. In most cases the links to the src.rpm and specifles are
invalid shortly after the review has been finished, thus rendering the
review ticket mostly useless. This doesn't however invalidate my
point. That issue has already been discussed here before, with no real
outcome, unfortunately.

-- 
Thomas Moschny <thomas.moschny at gmail.com>