BSDstats Statistics for Sept, 2007 ... 12 769 Hosts Reported In
Miguel Lopes Santos Ramos
miguel at anjos.strangled.net
Tue Oct 16 14:39:15 PDT 2007
> From: "Marc G. Fournier" <scrappy at freebsd.org>
> Actually, so far this month, they've gone up 0.3% (and its only middle of
> month) ... it does fluctuate from month to month ...
Just out of curiosity, on what criteria do you clean up your list? Maybe you're
cleaning it up too often, no?
> > I think it's because the bsdstats script is becoming a bit of a pain...
> > - The script keeps being changed and updated...
> Pardon? I have very rarely changed it ... I committed a change on Aug 29th to
> fix a problem with install.sh generating an error, and the last commit before
> that was on April 30th to clean up a bunch of portlint warnings ... prior to
> that, it was modified Dec 9th ... so there is about 4 months between updates
Yes, that is true. My memory deceived me. I still had the troubled initial
period of Aug to Dec, 2006 in my memory.
Clearly bsdstats is reaching stability. But consider that when you have a port
update, you don't know if it was the program which was changed or just the installer.
You have just commited another set of changes. I respect your decision of
commiting the changes.
On the other hand, I'm also not too keen on ports which require user
intervention on install, because of batch installs and upgrades.
Maybe that is not so crucial to a port like bsdstats, but still, wouldn't it
just be better to let the user go add bsdstats_enable to rc.conf?
I mean, we all have to do it for such basic things such as xdm or linux
compatibility. If every FreeBSD user is "required" to do such configuration
changes for those basic ports, why not for bsdstats?
Now, imagine you would agree with me, and went on changing pkg-install...
> ... we had several ppl reporting issues where they were using it on desktop's
> where the desktop was shut down over night, and would therefore *never* get
> reported ... if you are running a server, by all means, disable it in
> /etc/rc.conf ...
I understand that it is good to have it optionally on rc.conf.
> Oddly enough, when I did add the /etc/rc.conf option, it was asking if you
> wanted to enable it there or not ... now that you point it out, am looking into
> it, since it was meant to be optional ... in fact, will add a note to the
> question that states that enabling in /etc/rc.conf is meant for desktop/laptops
> that aren't necessarily 'always on', not for servers that are ... thanks for
> pointing this one out, I didn't realize ...
I saw you just corrected it.
> > While a lot of us understands the interest of this, and even takes the time to
> > take a look at how it's done, we also don't have the time to keep peting it
> > every time it changes behaviour. The first reaction might be uninstall the
> > port.
> Changing behaviour? Other then cleaning up some port code itself, and adding
> the /etc/rc.conf (which only affects things the first time installed), there
> has been no change in behaviour since ... December, from what I can tell, and
> the only change of behaviour then was to actually report non-configured devices
> that pciconf files ... so not 100% certain what you are talking about
> concerning 'changing behavioiur' ...
When we, users, see a port changing, we do not know if it is only the install
which changed or the program behaviour which changed.
Certainly the install has changed behaviour, and that is troublesome for
portupgrades done in batch.
Thank you very much, and I certainly don't want to dictate the behaviour
of bsdstats or its pkg-install script. It was just a complaint, because the idea
is excellent but it did annoy me a little bit at some particular times. It will
still annoy me a little bit because of the yes/no in pkg-install; maybe its me who
should be more criterious before doing my portupgrades...
More information about the freebsd-advocacy