[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Removing the number of supported installation methods



> Do we really *need* all of these?  From a test perspective, this does
> offer challenges.  I'm not listing all the command-line permutations
> where the boot media, install.img and package repositories are all in
> different locations.  Some suggestions ...
> 
>      1. Do we need support for *multiple* remote installation methods
>         (http, ftp, nfs, nfsiso) ... why not just HTTP (sure, libcurl
>         offers more, but HTTP would be documented and supported)? 
>      2. Are HD installs still valuable to users and customers of distros
>         that rely on the anaconda installer (preupgrade)? 
>      3. Are HD ISO installs still valuable?  
>      4. boot.iso, a boot CD, a boot DVD and a bootable Live image.  Is
>         there some way to combine these 4 boot images.  Do we need all
>         of them?  What about only providing a Live image only -- perhaps
>         more of a question for respective product teams (RHEL, Fedora).

Yes, we absolutely need to decrease the number of ways of booting and
finding installation media.  It's a completely unworkable situation.

My suggestion is we start by determining which methods are the highest
maintainence burden.  Which result in the most bugs?  Which are tested
the least?  Which involve the trickiest code?  Which have the least
valid use cases?  Then, we work on killing those methods.  Might as well
get the biggest bang for our code removal buck.

- Chris


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]