llvm case study: yum's handling of newly-converted noarch subpackages

Rahul Sundaram metherid at gmail.com
Wed Jun 9 05:03:47 UTC 2010

On 06/09/2010 10:24 AM, Kevin Kofler wrote:
> Good luck going through the hundreds of thousands of options of our hundreds 
> of packages and listing all the ones we don't support for whatever reason. By 
> the time such a list would be completed, it'd already be decades out of date 
> and thus completely useless!
> If you set a non-default option and your software doesn't work, the obvious 
> answer is "Don't Do That Then".

Generalizations such as this are not useful.  I didn't propose we go
through every option but if you knowingly include a broken option, it is
a different deal.  We will have to treat it on a case by case basis.  
While, it is obvious we would support the default set of packages with
their configuration better since they are the more tested path, some
non-default options are certainly going to be important for us to
support well especially in the core components.   Telling users to go
away when they report issues on them is hardly the right attitude.  


More information about the devel mailing list