On Mon, 15 Jul 2002 09:23:51 -0400 Charles A Edwards <[EMAIL PROTECTED]> said with temporary authority
> On Mon, 15 Jul 2002 02:25:56 -0500 > nDiScReEt <[EMAIL PROTECTED]> wrote: > > > On Sunday 14 July 2002 6:23 pm, James wrote: > > > OK! this may solve my problem as well... This matches my kernel > > > ... now the question comes.... Does this mean that Mandrake has to > > > build a new glibc for every older distro when 2.6 comes out? Or > > > do they just never receive the benefit of a new kernel? Same > > > thing for RH... > > > > Sadly, that looks to exactly be the case ...all systems will have to > > upgrade their glibc with it's respective kernel-header rpm in order > > to use the new 2.4.18 and up kernels. I don't have the official word > > so this is completely all conjecture based purely off of logic. > > > It is not and has never been the practice of any disto to provide > upgrades for previous releases. > This has always been the responsibility of the user. > Either by installation of a distro's 'newer' release or by building > from src the 'updated' app. > This applies to glibc, kernel, or any other app. > > What has been/is offered are 'updates', which during initial release > period provide for bug fixes but thereafter addresses only security > issues as with the 2.4 kernel and Apache "updates" for Mandrake 8.0. > > > Charles But this begs the question... If everyone is doing it does that mean it's the right way to go? Ask anyone at Enron and WorldCom... Again I still believe that without a means to upgrade a "live" or working system. (whether it's actively serving data or not, at the time of upgrade.) is a prudent business move. I've still got boxes running 7.1 for the simple reason... I can't afford the data migration time. Too much down time would be involved, So these systems continue to be less than optimal and probably will stay that way till the day the hardware dies. Failure to provide an upgrade path for the user without loosing data etc, is a very M$ concept. ( a recent "race" between my brother moving to XP and myself moving to 8.2 had me finished 2 days before him, and that was just to install software and reconfigure.... nothing else.) This alone is justification for doing it better. Why else would so many people still be running Linux with 2.0 kernels..... It works, and a path to migration is not available. (granted in many cases the 2.0 kernel has extreme advantages over the 2.4 but that's for discussion on kerneltrap not here.) Heck I don't know about anyone else but I think ... no I know I and many others would pay a monthly fee to have my, say 7.1 install upgraded a little every month... until by the time 7.2 came out I had 7.2... then continue on to 8.0 etc etc. The tools are there ... urpmi and MCC are way above anything RH etc has to offer. (OK my view is biased by preference ..) so the path is available. The nice part about this concept is it ensures that 1. Every user stays a user (they keep getting the latest and greatest. Moving to RH or SuSe away from MD would be relatively difficult.) 2 Users love the distro because data loss is minimal downtime is a reboot. (only if a new Kernel or glibc is installed.) 3. Mandrake gets a 1 to 1 "personal" relationship with their users 4. Income is greater (I haven't bought disks in 2 years for the boxes running 7.1 similar for other boxes running older versions.) 5. The need to release to increase is reduced because users are always getting improvements. So longer times between releases can occur and new users get a release that has be "field tested", and very stable. 6. It can be done online, or for those with modems... via cdrom monthly. (Hey it worked for FreeBSD till BSDi bought them.) Yes it does involve a change in strategy and wouldn't be an instant model. However I feel that it would present the possibility of cutting M$ off at the dollar in a way they can't legislate away. Not to mention a steady known income flow for the distro that does it. James
Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com