Slashdot Mirror
Fedora Core 3: Worth The Upgrade?
Chris writes "With new features such as SELinux, GNOME 2.8, KDE 3.3, Evolution 2.0, Remote Desktop, Helix Player, and of course Firefox, it may be worth your while to make the switch. At OSDir our screenshot tour of Fedora Core 3 takes you through boot, installation, desktop, taskbar, menus, configuration, and the new features of this new release. Our Core 3 screenshot tours have taken you through Test 1, 2, 3, and now the final release. Check it out."
AFAIK a minimal install only uses the first CD. A default workstation install uses three, but barely touches the last. I don't know what a default desktop install uses.
jh
As I found out the hard way over this past weekend, they left out all the java and java related rpms that FC2 had.
Are they using two different development teams for Fedora the way RedHat did for the x.1 and x.[02] releases?
I installed it on a fresh xeon 2.6ghz and I was abhorred at the slowdown. FC2 was a LOT faster than this is.
I'm not talking of booting into X and doing things in there. I'm talking just getting to a login prompt and attempting to sign on.
I'll go back to slackware before I load FC3 again
In which case do the HTTP install, and don't even download that much. I think the rescuecd can function for this purpose, and it's fairly small (about 80 meg).
jh
Frankly, a whole bunch of numbered image files does not make for a Fedora review. Personally, I can't even bother sorting through them all.
/dev only includes devices that actually exist in your system. This is kinda nice. e2labelas deprecated, as there's now a whole bunch of ways to uniquely refer to devices rather than just their label. This is good for people who hot plug a lot of devices.
/bin/sh, it runs as Old School Bourne shell.
I run Fedora Rawhide on my laptop. This would be the equivalent of say, Debian Unstable. So I have a good idea of what FC3 offers...
- Bluecurve theme finally covers everything.In particular, Firefox and OpenOffice look like every other KDE or Gnome app.
- If what I've seen in the RHEL 4 beta is the same for Fedora, partitioning now uses LVM by default. There's a new GUI LVM config tool called 'system-config-lvm' in Rawhide to provide the post-install disk resizing. Additionally, online resizing with ext3 should work and, if you use RHEL, be supported.
- Firefox and Thunderbird.
- SELinux turned on, including policies for locking down Apache, Bind, and NIS. A GUI config tool is provided for this.
- There's apparently improvements to yum which I'm not sure about. Personally, I'm a fan of up2date, which can use directories full of packages (without needing index files) as one of its sources.
- Udev.
- HelixPlayer is now included by default.
- Bash 3 - not much difference for me, apart from the new inbuilt range system that obsoletes the old 'seq' command. If you call it as
FC3 is really fast on my Athlon 1.4Ghz Thunderbird and it has really good font rendering but I experienced some hurdles:
My system has both EIDE devices and SCSI devices. If I use eg. my EIDE cdrom drive I cannot use my SCSI cdrw drive anymore as this system seems to use the ide-scsi emulation layer per default. The SCSI cdrw is only detected by Nautilus if I put a cd into it (I don't like these autostarters)
I tried to build ReZound http//rezound.sf.net/ but it failed to compile
Neither does Audacity
When compiling MPlayer it fails to build with GUI and it fails to play sound if you playback a video
These are problems which I don't have with my other SuSE system (on the same machine)
JAVA: I don't like to have gcj installed instead of a real JVM
MP3: none of the installed sound tools can play or record MP3 files
The eth0 device is automatically detected but the DSL configuration doesn't configure eth0 to be used with pppd. As a result the kernel tries to start eth0 but fails and the pppd connection starts afterwards. This unnecessarily slows down the boot process.
You have SELinux turned on. I've set mine to "Warn" until I understand it just a bit better. If you didn't turn it on, keep reading.
Once SELinux is disabled, run these in order:Should fix you up. The reason AFAICT is that the NVIDIA driver is not aware of udev, which FC3 now uses.
BTW, NVIDIA released a new driver the evening FC3 was released - go get that too : 1.0-6629
Soko
"Depression is merely anger without enthusiasm." - Anonymous
The new version does not solve the udev issue - you still have to run those three commands.
Soko
"Depression is merely anger without enthusiasm." - Anonymous
Judging by the 50 posts thus far, Red Hat/Fedora appears to have fallen out of favor with the averaging posting SlashDot reader. Nothing but a string of complaining, despite most being unfounded or flatly wrong.
Fedora Core 3 is a terrific GNU/Linux distribution. On one hand, it contains only Free software. No proprietary, patent protected, or closed source. Everything included is safe and the principled users of software can be at ease.
On the other hand, it is very polished. There are no dark corners of breakage, everything Just Works(TM). Network, video card, printing, CD burning, fonts, office applications, PDF viewing, email, file browsing, graphics, etc. All the little niggles of versions past (not just Red Hat either) been resolved to result in this super clean and functional distro.
As a Red Hat user since 5.0, Fedora Core 3 is the first version I feel is good enough for a non-geek Windows user to try. There won't be any surprises. Much of this is simply the development of GNOME 2.8, but Red Hat (ok, the Fedora Core team) has done an excellent job IMO of refining the base, too.
Now I'm sure posters can (and will) lament the downside. Fedora Core 3 will not be found perfect, featureful, fastest, most flexible, most standards compliant, most free, or the most usable. But across the board, FC3 is the best at fulfilling a balanced set of these qualities.
There is no need to use a SlashDot sig for SEO...
Actually you can do an http or ftp install from just the boot image which is about five megs. That's how I installed it. I don't see a reason to download several gigs of things I don't use, such as emacs.
I've been using FC3 for about one week now. Both of my external firewire drives (Maxtor) are working seamlessly with FC3. All three of my externals--these drives and my USB Flash drive--show up on the desktop (thus, fstab is working flawlessly). Real nice. I have a dual boot machine, with XP (NTFS). No problems with the install (I did a fresh install).
It's called SELinux, it was written byt the NSA, and even without it you already have pretty fine grained control over file permissions using groups and file permissions. What SELinux gives you is the ability to also restrict such things as network access, the right to fork, run execve, switch userid, etc. SELinux can grant these rights not only based on userid, but also on which program is run.
In the end, what this gives you is a system where, if a process using a properly configured SELinux has been taken over (0wn3d), it can't do anything other than screw up it's own job, unless it figures out how to fool SELinux.
Try out fish, the friendly interactive shell.
It's no problem however if you follow instructions on this page.
Actually Windows XP does include burning. Just copy files to drive and select burn (similar to OS X's implmentation). IMHO this is actually very intuitive a makes creating a simple data CD very easy. I am guessing that we will see a similar version of CD burning in the near future from some distros.
There is a CD-ROM on there.
I always thought that making your hard drives the masters and your CDs the slaves was the preferred arrangement? At least, that's what it said in one of the readmes in the kernel source last time I checked.
The next Cmdr Taco duplicate will be ready soon, but subscribers can beat the rush and see it early!
OK, I'm nieve. How do you do that - disconnect from a running display?
/dev/null, except that pts-inquieries won't fail (because you're talking to a real pts instead of a generic file). The biggest advantage, of course, is that an end user can disconnect or reconnect from the intermediate "screen" multiplexer (given the appropriate permissions).
Screen is not graphical but instead a pts (pseudo-terminal) multiplexer. Meaning a background process sits inbetween your pts port and your real terminal. "Screen" accepts all output (logs some fraction and dumps history-overflow), such that the underlying applications have no ability to determine whether there's a physical terminal at the other end or not. By accepting all input, it prevents applications from pausing once 2k of stdout has been queued but not sent/accepted. It's similar to redirection output to
Another feature of screen is similar to "virtual desktop". Since you have a multiplexer, "screen" allows a single physical terminal to switch between multiple pts channels.. So if you have a dial-up-modem (direct terminal, not TCP/IP), you can have dozens of different "windows" with different applications running in each (multiple vi windows, several command prompts, several log files, etc). If the modem hangs up, you dial back in, and type "screen -r", and you're back as if nothing had happened. You're alternative was to run all applications with "nohup myapp myargs" and if the modem hung up, then stdout would be redirected to a file.. This way you don't lose the output or have an interruption in say a slow compilation. But the problem is that you can't regain interactive access to a something like vi window. (Course, text editors have their own recovery capabilities).
So the original poster was trying to say that they wanted these incredibly valuable features in a graphical form. vnc and rdesktop allow a user that has their network connection broken to be reconnected without the underlying graphical applications ever being made aware of the interruption. With X and a static IP-address, there are time-out issues. And more commonly we move to different machines or different access points and thus necessarily can not recover a graphical session with X.
X was designed as client-server with state. It is this state that necessarily prevents it from acting like VNC or rdesktop. "screen", vnc, and rdesktop keeps it state on the machine with the running application. X keeps the state on the machine with the physical monitor/keyboard/mouse. I believe the original idea of this design decision was to distribute resources. The application server only performs tasks related to function, not display. Graphics becomes simply the ability to handle events and send graphical commands to a network access point. The terminal is then responsible for all resources related to interpreting graphical commands. This is similar to the postscript paradigm. postscript is a series of "logo" like commands (draw a box from this point to this point), and the printing resource determins how to render the fonts/color schemes, etc. Unlike postscript, however, X graphical commands aren't encapsulatable into a portable relocatable format since there is bidirectional communication going on.
Another particular of X is its peer-application structure. To run X, in addition to the terminal software and the physical applications, you need a font-server and a window manager. While this is great for pluggability (and even clusterability; running a single-threaded graphical program across 4 different machines), it necessarily provides greater latency for even simple tasks.
vnc merely adds a multiplexing layer to the back-end of X or windows, just like screen. So vnc necessarily adds a layer of overhead to the graphical process. More importantly there is an impeedence mismatch between the graphical transport of vnc and that of X. X is designed to send postscript-like graphical commands (draw a square of this size fi
-Michael