Philip Brown wrote: > Darren J Moffat wrote: >> Philip Brown wrote: >>> ... >>> [pkgadd] could be reworked further, to use the exact same zone-friendly >>> mechanisms planned for IPS. >>> (running pre/postinstall scripts through SMF, rather than raw, etc, etc) >>> That's not that much more work, compared to writing and debugging a >>> whole new packaging system. >> That doesn't help the problem that Sun faces that ISVs or other >> repositories like Blastwave or sunfreeware.com don't face: patching and >> the fact that the current patchadd/pkgadd are mostly unaware of each other. >> >> Sometimes it is good to start from a clean slate. I remember hearing a >> hole bunch of similar arguments against ZFS compared to UFS+SVM or >> VxFS+VxVM inside Sun when ZFS was around the same phase of development >> as IPS is now. Now look where ZFS is now! > > The big difference there, was that to achieve the goals that ZFS had set for > itself, you HAD to merge the disk layer, and the volume management layer. > It isnt possible to write a zfs equivalent, while preserving clean > separation between those layers. > > In contrast, when people take the time to actually list out specific goals > for IPS (rather than just handwavey terms like "innovative"), it is quite > possible to meet them, using the existing pkgadd paradigms. > > The patching issues are somewhat ugly, but primarily because of sun's > internal policies and proceedures of patch generation. > (and pkg grouping. some pkgs are too coarse).
How then do we resolve the scripting issue that SVR4 pkgadd has so that we can properly and safely support zones, diskless and cross architecture install/hosting ? While I freely admit the lack of scripting is causing me problems I fully understand why it is an issue. Sure we could use "process" to say that no SRV4 format package for Solaris can use scripting but unless the functionality was removed the problem still exists. Given that the SRV4 packaging tools are open source, if you really believe it can be fixed you are able to do so. I think it would be good to see that done even with IPS - competition is good. Just don't expect that the IPS team are going to do that they have their project to do. -- Darren J Moffat _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
