On Mon, 18 May 1998, Igor Grobman wrote:
> Well, you could compare the md5sums, but that's not what you want to do... > In > your case, it's better not to include the conffile in the package, but > instead > create it in the postinst. You should probably separate the config-file > creation part into a separate script that the user can run any time s/he > feels > like. This is what gpm, sendmail and many others do. > > BTW, modifying conffiles in maintainer scripts is against policy (or > packaging > manual guidelines which should be policy). > This sounds like it could get complicated. Actually now that I look at it, there is only one piece of information a user needs to make pppupd operational--a hostname to ping. I can make this localhost and just tell the user to change it to something more useful. > Well, I can't think of a way to see if ppp is truly configured, not just > successfully installed (this is what you will check for with your method). Good point. > Just making pppupd depend on ppp will make sure that ppp will be installed > before pppupd is configured. Yes it depends on ppp. > I think that in this case, you have to assume > that the user configured ppp already and/or do what you are already do: warn > the user that the package won't work without ppp being properly configured. > The problem with the warning is that on a long deselect run it will whizz by to fast for anyone to see it. So what I do is ask if ppp is properly configured and actually make the user type Y or N. The configuration only continues if they say yes. I realize this could be considered quite obnoxious too but I think it is very important pppupd not run until ppp is properly configured. Still If it is considered really really obnoxious I'd get rid of it. -- Jaldhar H. Vyas <[EMAIL PROTECTED]> -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
