Efficiency & correctness for port version upgrades

[Doug Barton]

On 07/31/10 05:00, David Wolfskill wrote:
> I have been using portmaster for updating the installed ports on
> may machines -- thanks, Doug!

My pleasure. :)

> I'm in the habit of updating on a daily basis -- directly after the
> "smoke-test" boot of a newly-rebuilt stable/7 (before I go on to
> build stable/8).
> 
> This morning, I noticed the (20100730) entry in UPDATING on www/neon28
> -> www/neon29, so I made sure to complete that before doing the
> "svn update" on my stable/8 working directory.  [This is one place
> where having the revision control system in ports, vs. base, is a
> bit of a nuisance -- but that's off-topic for this note.]
> 
> The instructions given for portmaster were:
> 
>   Portmaster users:
>         portmaster -o www/neon29 www/neon28
>         portmaster -r neon29-
> 
> which is fine, in that it works.
> 
> However:  It also causes www/neon29 to be built twice; I cannot
> help but question the extent to which this is actually desired.

One disadvantage (or is it an advantage? I'm never sure) of the fact
that portmaster does not maintain state between runs is behavior like
you're seeing here. To answer your question, assuming that everything
went well with the -o command, there is no reason that portmaster would
have to build the port again. The -r option needs an up to date
+REQUIRED_BY, which is a side effect of (re)building it; but the above
could just as easily be written:

portmaster -o www/neon29 www/neon28
portmaster --check-depends
portmaster `cat /var/db/pkg/neon29-0.29.3_2/+REQUIRED_BY`

I often get the request for this feature (a -r that doesn't rebuild the
port) but have never found the time to work on it. I'll try to get
around to it soonish.


Doug

-- 

	Improve the effectiveness of your Internet presence with
	a domain name makeover!    http://SupersetSolutions.com/

	Computers are useless. They can only give you answers.
			-- Pablo Picasso