Luca Berra wrote: > On Sat, Jun 07, 2003 at 02:12:47PM +0200, [EMAIL PROTECTED] wrote: > > I DO agree with you in that the user should be able to check the > checkbox, if you leave me the possibility of modifying what the checkbox > does, that is.
Well, you would then probably want to be able to modify what it says next to the checkbox too ... > I DO NOT agree that autocreate or auto-(assume that the luser wants this > so do it silently)-anything is user friendly. > > I would personally like a single framework for configuration, but all > frameworks i found in 13 years working on unix systems do suck far more > than vi. It's no problem though if they work well with vi (ie not removing comments, not storing configuration in some other meta-configuration file so your changes are overwritten, for example). > I am not saying it cannot be done, what i would like, but i don't know > how to achieve is having a configuration framework that > a) doesn't hide details leaving the user in need of using vi > b) does not make assumptions > c) has decent sanity checks (drakconnect once for obscure reason addedd > ip 10.0.0.1 for my workstation at top of /etc/hosts, i took hours to > understand why random things were malfunctioning) > d) has proper configuration history/rollback features This is one I haven't asked about, but was thinking about on Friday, this is necessary, and not too difficult ... > e) lets the admin configure it with cutomized host or site wide > policies (these should be enough to replace assumptions). The site-wide bit is going to be more complex, MS does it via Active Directory (Group Policy objects stored in the OU or the "site" configuration). > f) is able to adapt to changes in configuration options of the single > subsystem. Provision has been made, IMHO, the "template" (as it is called in libconf) should be owned by the package, but in libconf it seems there will be a template owned by libconf, but it should take into account an alternative template file. > I never asked for another web frontend, i am just stating that their > existance is caused by repeated failures of creating a decent framework, > or agreing on the need of one, we have tons of linux distros each with > it's own config framework, plus a good number of non distro related > tools. Well, libconf is possibly also going to be used by Gentoo? Anyway, it seems the most complete framework so far (and I was subscribed to http://unixconfig.sourceforge.net, which was aimed at solving this problem, and at least had discussions covering all the frameworks available at the time, but never go off the ground). > M$ did not have this problem, and did not even have to deal with 100+ > different configuration syntaxes, and 10+ different gui frameworks, > hence the better result. There isn't really an issue dealing with multiple gui frameworks, if there is one configuration backed, which interfaces to the 100+ configuration syntaxes, and provides hints to the gui (on what needs to be configured, descriptions, etc etc), as any competitive gui framework will find themselves needing to make a frontend (which should need no knowledge of any service, only of the backend). This is more or less what libconf is aiming at ... but at present I don't think there are enough people working on it to complete it for 9.2 (by complete, I mean covering all the services available in main). With community involvement from people who can hack a regex and know one service pretty well, it would be feasible though. Regards, Buchan