Weekly Edition
Return to the Press page
| |||||||||||
Linux Creator Calls Backporting 'Good Thing' (InternetNews.com)
InternetNews.com talks with
Linus Torvalds and others about backporting. "Torvalds comments,
in an e-mail interview with internetnews.com, came after SUSE'S CTO,
Juergen Geck, told an audience at the Real World Linux Conference in
Toronto that Red Hat's practice of backporting features from the 2.6 kernel
into the 2.4 Kernel is a "bad thing" because it interferes with
standardization of the open source operating system."
(Log in to post comments)
Linux Creator Calls Backporting 'Good Thing' (InternetNews.com) Posted Apr 21, 2004 3:24 UTC (Wed) by marduk (subscriber, #3831) [Link] I don't know about the kernel per se, but one thing I really appreciated about Red Hat is that when there's a security advisory for an application, they will, when possible, backport the fix instead of forcing you to upgrade to version n+1. That's a really great thing to have in a production environment when you don't need more features, just less bugs.
Linux Creator Calls Backporting 'Good Thing' (InternetNews.com) Posted Apr 21, 2004 7:23 UTC (Wed) by rsidd (subscriber, #2582) [Link] Debian stable does that too for packages (that's why the packages all seem antique). However, the article wasn't referring to porting bugfixes from 2.6 to 2.4, but to porting entire new features.
Linux Creator Calls Backporting 'Good Thing' (InternetNews.com) Posted Apr 21, 2004 9:07 UTC (Wed) by pointwood (subscriber, #2814) [Link] That's not the policy of Fedora though, AFAIK.
Linux Creator Calls Backporting 'Good Thing' (InternetNews.com) Posted Apr 21, 2004 12:24 UTC (Wed) by clugstj (subscriber, #4020) [Link] SuSE's CTO would complain about RedHat doing this because it means thatRedHat's distribution then has more cool new features than SuSE's.
Linux Creator Calls Backporting 'Good Thing' (InternetNews.com) Posted Apr 21, 2004 13:51 UTC (Wed) by QuisUtDeus (guest, #14854) [Link] An even better result of the back-porting is that an expanded set of eyes is looking at the code and trying to figure it out for inclusion in the older kernel framework. This can lead to the discovery of subtle bugs and possible exploits.A very good thing.
Wasted time for driver maintainers Posted Apr 21, 2004 15:42 UTC (Wed) by proski (subscriber, #104) [Link] On the other hand, maintainers of separately distributed drivers get tons of e-mails about the driver not compiling against the vendor kernel. Red Hat put free_netdev() in their 2.4.22 kernel, while the official kernels started using it in 2.4.23. Some other distribution decided to backport work queues, which required workarounds to prevent conflicts between two sets of compatibility code.I'm not against backporting, but it has its price.
Wasted time for driver maintainers Posted Apr 21, 2004 17:53 UTC (Wed) by QuisUtDeus (guest, #14854) [Link] Documenting the abnormal assumptions of a back-port would be polite: "This backport requires this other patch," etc.
Documented in the src.rpm Posted Apr 21, 2004 18:33 UTC (Wed) by AnswerGuy (subscriber, #1256) [Link] ... the documentation for Red Hat vendor changes is the .spec file in the SRPM.I'm not going to comment in either way on the "good" vs. "bad" question because I see it as a set of trade-offs. JimD
|
|||||||||||