[PATCH] Revised patch for KUser

[Oswald Buddenhagen]

On Tue, Jun 08, 2004 at 03:54:58PM +0200, André Wöbbeking wrote:
> On Tuesday 08 June 2004 00:50, Oswald Buddenhagen wrote:
> > On Mon, Jun 07, 2004 at 09:40:00PM +0200, Jan Schäfer wrote:
> >
> > > +  QValueList<KUserGroup> groups() const;
> > > +  QValueList<KUser> users() const;
> >
> > now that is look at this, i'm not sure whether it is a good idea to
> > return lists of KUser[Group], as this may be excessively expensive.
> 
> QValueList is implicit shared with copy on write semantic, isn't it?
> 
that's not the point. i'm talking about constructing several, possibly
doozens of user/group objects even though we might be interested in
their names only.

> > maybe it would be better to return string lists and let the user
> > create the respective objects himself if necessary.
> 
> in every place where this is used?
> 
examples?
without hard evidence it seems reasonable to take the less expensive
route.

> > not sure about all{Users|Groups} ... without the auto-construction
> > they are trivially done with the posix functions, and i don't think
> > we want plain wrappers in the api?
> 
> I think this is the idea behind this class a C++ wrapper for the Posix
> api. Not all of us are so familiar with the Posix api as you.
> 
that's a pretty nonsensical argument. if you have no clue about the
unix/posix user/group concept, this class won't help you, either. and if
you have, finding the relevant docs is trivial. heck, we could/should
even reference them in our doc.
apart from that, this paragraph was considering the exact opposite of
the previous one. you can't dislike both. ;)

greetings

-- 
Hi! I'm a .signature virus! Copy me into your ~/.signature, please!
--
Chaos, panic, and disorder - my work here is done.