<div dir="ltr">Hi<br><div><div class="gmail_extra"><br><br><div class="gmail_quote">On Wed, Jan 22, 2014 at 5:32 PM, Miloslav Trmač <span dir="ltr"></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<br>
1) No, that's not actually true. The idea of an all-encompassing<br>
Linux distribution is fairly unique in the world of software;<br>
everywhere else you have users installing software from various<br>
sources and being used to expect support (or not expect support) from<br>
the individual vendors of the individual products.<br></blockquote><div><br></div><div>I don't think is a fundamental disagreement here. Linux distributions have set expectations for users that is different from say Apple or Microsoft but given that we already set those expectations, we have to figure out how to reset it if we enable access to a bunch of software that we don't maintain or have direct control over. Otherwise, users are going to assume that we will be making some reasonable efforts to get the combinations tested even if we call it unsupported c.f kernel-unsupported in past RHEL releases.<br>
<br><br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
2) That said, yes, a platform where the user anecdotally doesn't<br>
expect installing software to be problematic is going to be more<br>
popular than a platform where problems arise. This problem also has a<br>
well-established solution by now:<br>
2a: The operating system implementation ships APIs and documentation<br>
that make it easier to write software that doesn't break<br>
2b: For software that does the wrong thing and breaks, the operating<br>
system adds a special case to make the software work anyway; blaming a<br>
different vendor and keeping a clean implementation may be<br>
intellectually satisfying but doesn't help the user get their work<br>
done, which is what really matters.<br></blockquote><div><br></div><div>So, like I said, pretty deep changes in the way we do things in Fedora. If you can achieve ABI stability and provide a distribution neutral method of deploying software, you will reduce the overhead for vendors who do want to provide something for Fedora. Linux distributions have used the repository model primarily to workaround the lack of platform stability and we have only been able to do that because we can rebuild everything in the repository anytime an ABI bump happens. If we are going to change that, we will have to continue shipping things like HAL as an example, because Amazon instant video depends on it and currently users to have to jump through hoops to get it working again. Enabling direct access is the very last step of that process. We cannot just mimic the app store model without handling all this.<br>
<br></div><div>Rahul<br></div><div> </div></div></div></div></div>