Ports with GUI configs

Chuck Robey chuckr at chuckr.org
Thu Nov 15 13:02:19 PST 2007


RW wrote:
> On Thu, 15 Nov 2007 14:55:02 -0500
> Chuck Robey <chuckr at chuckr.org> wrote:
> 
> 
>> Are you seriously saying that a decision regarding what ports are to
>> be installed should be made after they are installed?  If you have
>> 10,000 ports installed, you obviously have no need whatever to make
>> any decision at all.   Whether or not they are outdated is utterly 
>> irrelevant, because if they're installed, it may be inferred that you 
>> wanted them.  It's the decision whether to install them or not that 
>> we're talking about.
> 
> 
> 
> What you don't appear to understand is that the Option Framework allows
> a user to recursively set options for ports *before* they are installed.
> So to configure the whole of Gnome, you can simply do this:
> 
> # cd /usr/port/x11/gnome2 && make config-recursive
> 
> The reason I mentioned the script is that upgrades are the only part of
> the process that isn't directly supported by the ports system, you need
> something to catch the ports that have changed options, or you may
> waste time. This requires a script, but new installs are completely
> trivial.


I've already deleted the message that kicked me off, but it looked to me 
that you were talking about the 10,000 ports I was talking about, and 
that meant you were referring to new installs, not upgrades.  If it were 
me, I would think that upgrades should probably follow the same path as 
the original install, no?  In some small number of cases, there would be 
brand new options that would not be possible to predict from the 
decisions already taken for the orignal system, but that would be the 
exception, not the rule.

Regardless, as an unintended side effect of the system I'm talking 
about, such items would be automatically taken care of.  The only 
recurring task would be, as new options find themselves required, users 
would be asked to register the setting for a new keyword.  This would 
probably mean something on the order of maybe one or two new words a 
month to decide on, something that would hardly be a worry.

I do agree, the system I'm talking about, if I was trying to justify it 
only on the basis of upgrades alone, would not be justified.  Sort of 
like the tail wagging the dog, too much work for too little gain, but as 
a nice side effect, it's acceptable.

BUT if you were talking only about upgrades, then I kinda think, 
personally, that you probably should instantiated a new thread, not used 
this one.  Hmm?


More information about the freebsd-questions mailing list