> On Thu 2003 Oct 23 at 13:01, "Buchan Milne" <[EMAIL PROTECTED]> wrote,
>
> in part:
> > ...
> > But, I don't think anyone on this list qualifies as a newbie.
> > People on this list should be testing at least half the betas
> > on at least one machine, and/or running cooker full-time on a
> > box (if you don't, you're most probably a lurker ;-)).

 I'd love to run cooker and am more than competent to test it but there are 
some minor issues that I think you'll find common among most of the lurkers 
like me.

1) I only have one computer (University Student) it's an oldish system (k6-2 
500Mhz) that runs everything alright if abit slowly. But it's my only 
computer and is used daily. I'd like to run cooker on it but, I need it to be 
useable I can not risk having it go down for a day or two due to a bug in 
cooker or one of the apps I use being unusable for a day or two while bugs 
are sorted out. I know there are ways to install multiply versions of linux 
on the same machine and this would work quite well having the current stable 
version on one partion and cooker on the other would allow cooker to be used 
all the time while giving me a method to use my computer if the cooker 
install stuffed up due to bugs. How ever I don't know enough about how to set 
this up and  have never been interested enough to find or work it out due to 
the second  and Major problem.

2) Bandwidth or datacap/international data costs. This I think is the biggest 
problem many Lurkers I think you'll find are using either 33.6K, 56k modem's 
or Broad band or semi broad band (eg I've 128Kbps ADSL) with a datacap eg 
500MB-5GB a month of international traffic. 
Those with standard Modems simply can not keep up with the rate at which 
changes are made with new versions coming out before they've even finished 
downloading the previous ones. While those with the Broadband and datacaps 
either simply can't afford to download (eg only 500MB international traffic) 
all the changes or are unsure about how much data updating is going to take 
and can't risk going over their cap, As over the cap people tend to get 
charged rather high prices for every extra MB. In NZ it's about $0.20 a MB 
which if you go over the cap by a fair amount say a couple of GB works out to 
a fair amount of money.

3) Bugzilla It's big it's and it's slow and there nearly allways seems to be 
some sort of problem with it.


There are ways around these problems and working them out I think would 
significantly increase the number of testers. Some of these Ideas would help 
quite alot I think.

1) Develop a guide or utility for installing cooker easily along side another 
Distro or release. Or simply away to install two separate copies of the 
current release on too separate partions, urpmi can then update it easily 
enough.

2) Provide a guide or a tool to use rsync to maintain a mirror of those 
packages you've currently got installed using cooker. Most people will not 
install every thing on the various cooker tree's. If urpmi or some other tool 
could make a mirror of just those packages you've installed from cooker on 
the drive and maintain it using rsync alot of bandwidth and time could be 
saved making it easier for those with limited datacaps or bandwidth to 
maintain their cooker install.

3) Have weekly, biweekly or monthly branches. Demi freezes. Again those with 
limited bandwidth or data limits would find this easier to work with. If they 
could download a branch of cooker that would remain stable for a couple of 
weeks or months they'd be able to heavily test out cooker finding valid bug 
with out having to update daily combined with an rsync mirror util this would 
allow the amount of bandwidth required to keep up with a usefull version of 
cooker reasonable. This would be especially usefull for those with 56k 
connections. As they would not have to worry about version changes when 
updating using this source. and would only have to update once or twice a 
month. This would also probably help to find and ID bugs better with multiply 
users using exactly the same versions of cooker bug's could be Id and tracked 
much better. It would probably also allow more in depth testing of new 
feature allowing more bug's to be found .As if they found a bug unless it 
made the system unusable they'd stay with that package and continue to test 
it looking for other bugs rather than straight away downloading a new package 
that while fixing the current bug with a patch introduced half a dozen new 
patches and bugs.

4) Test Install ISO's Small 100-200MB isos that contained just a minimum 
install and A desktop allowing detailed testing of CORE packages and 
installation. eg KDE + the main apps oo.org gimp etc.
Produced monthly or biweekly again updateably by rsync.
Most likely 2 or three would be needed.
KDE + OO.org + Gimp + Mozilla.
Gnome + OO.org + Gimp + Mozilla.
And say.
Icewm + Server apps.

NO "development packages" NO "extra libs" just a small neat Desktop install 
the kind that most USERS use which as far as I'm concerned should work 
perfectly in all releases.
People can like with RANDOM-APP-XXX not working but they'll never support a 
distro can't get the install + basic desktop working.
This could also be used to test out the upgrade feature properly.

5) A small "fast easy" to use client side front end for Bugzilla! With a nice 
simply understandable name and Description so that non developers testing can 
understand it.
eg. BUG-Reporter >> Used to report bugs to Mandrake. 
Possibly with some scanning ability to grab system details (with the users 
permission of course) about the computer from /proc/ and package versions for 
any apps that the one reported to have a bug in has listed as a dependency.

Chad

Reply via email to