Nicolas Williams wrote: >On Tue, Jan 09, 2007 at 06:25:03PM +0100, Darren Reed wrote: > > >>>But why should this 'fetchfile' utility be seen as SMF-specific? >>> >>> >>Hmm, you've got me thinking, so let me think about this... >>Let's see, the smf_fetchfile(): >>(1) needs to ensure the pathname returned is part of the filesystem >>(2) should preferentially cache said file if it isn't local >>(3) potentially be able to return a local filename for a cached copy if >>the remote one cannot be checked >> >>Should it be SMF specific? No, I can't see why it needs to be. >> >> > >Me either. > > > >>But the primary focus of the idea is to make it easier to write >>SMF manifest scripts, so it needs to either be a program that >>lives in /bin or a shell library function. I'm not aware of any >>general purpose shell library files that you can "source" from >>your own shell script on Solaris, so going that path doesn't >>seem like an option. >> >> > >You could write a shell script around wget, or just use wget directly... > >
A shell script around wget would be a start, yes. >>In implementing (2) and (3) above, it may be desirable for the >>smf_fetchfile() to use a "private" directory/cache. >> >> > >That's another story, and there's security issues to consider in writing >a local cache server. > > My thoughts on this would be that a local cache server is a bridge too far. It's too much work/effort/complicated. Darren
