On Saturday 23 August 2003 00:22, Jeffrey Smelser wrote:
> OK, maybe its just me and I am a control freak.. (hell, my wife says I can
> be.. :) ).. How can anyone just write a script to emerge world all the time
> and expect the system to just be ok?? Maybe its because of my systems being
> here so long.  I have used gentoo since very early in the process as I
> followed Robbins around with his how to articles and was there when he
> decided to create this gentoo.

I am one of the ones who does automatic updates, insane CFLAGS, runs 
2.6.0-test3-mm3, gcc 3.3 and ~x86. As I said previously in the e-mail I'm 
sure prompted this one ;-), I've never had anything break so bad that the 
system's been unuable. The way I see it, someone's got to find the (almost 
always) small bugs that are missed by the developers and I'm happy to do 
that. I'll never have the time to do any real development for Gentoo so I'm 
just doing perpetual beta testing. I did a lot of beta-testing with the MS 
lines so I'm used to much worse bugs! ;-)

> My problem with it is this, I have rendered things useless, normally
> because the update is a considerable change and the config files change..
> Being I am sitting right there during every update I do, I just fix it and
> move on.. but how can you script this out? How do you know if samba is
> going to send you that fix that you must change something in your config
> file??

The e-mail my script sends me tells me what packages were installed and at 
what point and why it failed if there was a problem. The first thing I do in 
the morning is check to see if any updates were made and then run etc-update 
to check for any modifications that may need to be made. If I was running a 
server or something like that, there's no way I'd do it but, as it's just a 
home system, nothing is really running when I'm not using it so it's 
relatively safe to update.

> The Perl thing is a great example of an update I have had recently that if
> I wasn't sitting there when I did a emerge -udp, I would not have noticed
> the new perl broke several of my perl packages... Then telling me I needed
> to uninstall them first, install perl, then put them back... Then you got
> people on here asking why they have these perl packages that want to be
> backed out since they don't know what's going on.

The perl thing is an example of something that didn't halted my automatic 
updates. There was a couple of packages installed that blocked the perl 
update, so I had to sort it out manually the next day.

To sum up, if you want absolutely no breakage then Dhruba, Collins and Spider 
have the right ideas - especially Collins with his "wait two weeks" policy. 
Actually Spider broke his test box recently and did it all by himself - but 
that was a hard drive failure! If you can accept a small amount of breakage 
and have the time and intellect to find the solutions, then automatic 
updating is not such a bad thing.

Regards,
Jason

P.S. My reference to intellect does not imply that the aforementioned people 
are lacking in that area - there responses confirm that they are not. My 
point is that you're average user migrating from RedHat or possibly even 
Debian would probably not have the troubleshooting skills necessary to cope 
with the amount of breakage that does happen.

--
[EMAIL PROTECTED] mailing list

Reply via email to