Why does portsdb -Uu run so long?
Donald J. O'Neill
duncan.fbsd at gmail.com
Sat Feb 4 20:01:07 PST 2006
On Saturday 04 February 2006 16:56, Xn Nooby wrote:
> By the looks of it when you cvsup you get everything (src-all,
> > ports-all, etc) all at once. I think it might be better if you split
> > that into two sup-files where you would have one for the system,
> > src-all, and the other one for ports. This way you don't have to
> > rebuild the system every time you update your ports, this also works
> > the other way around. Once a branch is cut and declared -STABLE the
> > libraries used to make your programs work are rarely changed, If it
> > does change they will tell you in /usr/src/UPDATING. For the sake of
> > troubleshooting it helps if you don't change everything all at once.
> I thought that maybe by changing everything at once, I would avoid
> mismatched libraries.
> Someone should write a book on all this stuff, and explain it thoroughly,
> with various case examples. When I use the old slow way, I never get an
> error - when I use portsnap, I do. This makes me inclined to never use
> portsnap, regardless of how fast it is.
Xn, it appears to me that you are doing a lot off work for not much gain.
There's no reason to do a buildworld sequence everytime you upgrade the ports
tree. There's no reason to do a massive portupgrade just because you did a
buildworld sequence. I you are running a release version (the same goes for a
security release), the only time you should have to do a buildworld sequence
is if there's been a security update, and that's pretty much it. If you're
running a stable version, you may run a buildworld sequence more often, but
Now, if you want to run portsdb -Uu, that's up to you; but by the time you get
done with the sequence you use for updating your ports tree and get into
upgrading with portugrade -arR, I'll be done and using the system. Portsnap
doesn't have errors. 'Make fetch index' doesn't have errors. Not too long
ago, portsdb had a problem with ruby and a great cry went up in userland. And
much advise was given by the users who didn't have the problem to the users
who did, most of it false. A few good workarounds came out of it though.
I want to ask you: how long does it take you to cvsup your ports, run 'portsdb
-Uu', and finish with 'portversion -l "<" '? To run 'portsnap fetch update',
then 'portversion -v | grep needs', it took less then 55 seconds and I was
off upgrading ports. The procedure I used had no errors.
Both ways of upgrading work. Neither way will tell you about the conflict
between pilot-link and libmal. You're going to have to find out about during
an upgrade or, or wait and read about it on the list. So you can't be talking
about that as a problem with portsnap. Just what was the problem you had with
More information about the freebsd-questions