Sid Boyce wrote:
http://www.mandrake.tips.4.free.fr/opensuse10.3.review.html Mentions of "for the uninitiated" sounds about right. To sidestep any problems with the DVD, I've installed from Factory. Looks like the DVD should have been delayed until it was compared with Factory for stability. I think the reviewer attempted to be fair (he points to the URL to show where he is coming from). Perhaps some pertinent pointers for 11.0. Regards Sid.
http://www.mandrake.tips.4.free.fr/opensuse10.3.review.html should be MUST READING for management at Novell and at openSuSE.org. It is eminently fair and for the most part, accurate. I think any inaccuracies could be attributable to unfamiliarity rather than bias. If this reviewer hadn't been what I would call a 'Power-user', and had instead been someone converting from Windows, not only would they probably not have had a successful install, but would have an article that Microsoft would probably have printed in full page ads throughout the world. As it is, it should wake up Senior Management in SuSE.
regards Richard
Richard Creighton wrote:
eminently fair and for the most part, accurate. I think any inaccuracies could be attributable to unfamiliarity rather than bias. If this reviewer hadn't been what I would call a 'Power-user', and had instead been someone converting from Windows, not only would they probably not have had a successful install
completely wrong.
because he is a power user, he have four linux on his laptop. It's already good if he could install at all. He don't say what answers he give to the questions, but, for example, is glad to have Yast asking if he wants to keep old users. Do you think he could have such question on a fresh install? I beg no. He is making an update. and what was the old Linux? who knows?
so may be he got wireless working because it was working on the old install and so the firmware was at hand (how could he have it if not?) and x don't working because the old install was not completely compatible with the old one.
any intelligent review must be done on a fresh (not new) computer if it wants to be other than individual experience
this is the install of a power user upgrading an old install.
jdd
jdd wrote:
Richard Creighton wrote:
eminently fair and for the most part, accurate. I think any inaccuracies could be attributable to unfamiliarity rather than bias. If this reviewer hadn't been what I would call a 'Power-user', and had instead been someone converting from Windows, not only would they probably not have had a successful install
completely wrong.
because he is a power user, he have four linux on his laptop. It's already good if he could install at all. He don't say what answers he give to the questions, but, for example, is glad to have Yast asking if he wants to keep old users. Do you think he could have such question on a fresh install? I beg no. He is making an update. and what was the old Linux? who knows?
Balderdash! Are you saying that because he is experienced and that he has other distros or OS's on his system that he can't write an objective review and if he wants to upgrade his system from an older version instead of doing a 'fresh install', does that make his review less valuable. Few people I know want to erase months or years of configuration effort by the equivilent of FORMAT C: just to get a 'fresh install'. I have communicated with many from around the world that simply wanted to upgrade from 10.2 to 10.3 and the system failed miserably. Why should that not be reported if it is a fact? To insist that the only VALID test of an operating system or distro like 10.3 is to wipe your disk and start fresh in order to avoid problems is ludercrous and utterly unrealistic. It is lucky that he was a 'power user' or the review would likely have been a LOT less to your obviously biased liking.
This group was asked earlier by management of openSuSE.org to try out other distros and report back our findings. I can't think of a better report than a user of another distro giving a reasonably fair evaluation of *our* product. If it has warts, we should be willing to admit it and do something constructive about it rather than burying our collective heads in the sand and saying '...but he is a power user on a special machine' and ignoring what he had to say.
so may be he got wireless working because it was working on the old install and so the firmware was at hand (how could he have it if not?) and x don't working because the old install was not completely compatible with the old one.
He did give credit where credit was due. He only said that in many cases, Yast, is not very intuitive and that also is a fact. It is a dammed good program, but that doesn't mean it is perfect and can't stand improvement.
any intelligent review must be done on a fresh (not new) computer if it wants to be other than individual experience
Balderdash! Again! I think doing it on a so-called 'fresh' machine is very Microsoft in its' mentality. Who did you say you work for? Real life dictates that most installs will be upgrades not fresh installs and tests that ignore that fact are themselves deficient. So, if as you say, he did an upgrade, (which I didn't read from his report but might have missed), that makes the test even more valid, not less.
this is the install of a power user upgrading an old install.
So, why does SuSE give the option to UPGRADE....if it isn't a valid method of getting the distro onto ones' disk, then it should simply erase everything and start fresh. That should make you happy and you can write your own review. Be sure and complain about the loss of all your favorite programs and settings and the amount of time it takes to relocate the programs and set them up again from scratch. It is your machine, after all, but for me, I want to know if I can still have success upgrading or if I can have dual or quadruple boot to other distros or even Windoze. Any review that leaves out that info is less valuable IMO. And lest we forget, right up front, he said that it had been a long time since he had used SuSE and he felt is was time to give it another try....which I think he did. So, again, SuSE management *should* read his review because it is not written from one of the lemmings, but from someone actually spending a bit of time and trouble evaluating the competition.
Richard
jdd
Richard Creighton wrote:
Balderdash! Are you saying that because he is experienced and that he has other distros or OS's on his system that he can't write an objective review
he can't say "I don't recommand this for a beginner, because a beginner won't have 4 linuxes on his machine.
I *never* make an upgrade, and this with any distro (I tried many, specially mandrake or debian for years) because there are always a lot of problems.
and if he wants to upgrade his system from an older version
instead of doing a 'fresh install', does that make his review less valuable.
of course. this makes he's computer very special (as any of os, probably). This make a valuable *personal* experience, but nothing that can be reproduced
Few people I know want to erase months or years of
configuration effort by the equivilent of FORMAT C:
format c: in linux???
just to get a 'fresh
install'.
keep you home as I do and use a new partition for new linux... and things go mostly good
I have communicated with many from around the world that
simply wanted to upgrade from 10.2 to 10.3 and the system failed miserably. Why should that not be reported if it is a fact?
the problem is that is this review, one have to *guess* what tha author did.
insist that the only VALID test of an operating system or distro like 10.3 is to wipe your disk and start fresh in order to avoid problems is ludercrous and utterly unrealistic.
but this is true for any linux I ever could use... too many things change from one distro to an other (you can't either upgrade from xp to vista or windows 2000 to xp)
collective heads in the sand and saying '...but he is a power user on a special machine' and ignoring what he had to say.
I know of users able to go to a car trash and build a marvelous car from there, it's not what most people do.
the author expereience is good, but he have visibly no experience of scientific protocol:
* use known starting point, always the same, completely described * apply object of this system * report * do it again with any object you want to compare
cases, Yast, is not very intuitive and that also is a fact.
what mean 'intuitive'??? in most if not all the cases, intuitive mean "acostumed with". Last saturday I had to debug kubuntu and mandriva 8 on friends computers, and I can say they interface is very unintuitive for me. Ihopenit become intuitive for regular users (but it was not for my friends neither...)
It is a
dammed good program, but that doesn't mean it is perfect and can't stand improvement.
of course
Real life dictates that most installs will be upgrades not fresh
most people uses preinstalled computers and Linux (any linux) will always have problems until preinstalled... the reporter said also that yast is very good as keeping account of other os on the disk, this is the most important
installs and tests that ignore that fact are themselves deficient. So, if as you say, he did an upgrade, (which I didn't read from his report but might have missed)
no, he didn't saif that. One had to guess from he's reports
can write your own review. Be sure and complain about the loss of all your favorite programs and settings and the amount of time it takes to relocate the programs and set them up again from scratch.
? I did this two weeks ago. from yast this costed me one afternoon (use to be one week for W) when I don't want to loose any setting I simply don't change distro :-))
*should* read his review because it is not written from one of the lemmings, but from someone actually spending a bit of time and trouble evaluating the competition.
it's a valuable personal but very badly reported experience.
for example, how do you want to compare things not knowing what coputer he used? here what I did when buying my new laptop next week
http://new.dodin.org/wiki/pmwiki.php?n=Main.Acer9410z
with this I could report.
and I'm also a beta tester...
jdd
jdd wrote:
Richard Creighton wrote:
Balderdash! Are you saying that because he is experienced and that he has other distros or OS's on his system that he can't write an objective review
he can't say "I don't recommand this for a beginner, because a beginner won't have 4 linuxes on his machine.
jdd,
I just thought I'd enclose a direct E-Mail from the person that wrote the review. I think it explains or addresses the issues you raised fairly well. At least, I think it amplifies some of the points you questioned in some of your earlier posts.
Quote:
Dear Richard, you don't know me, but you've read my work at: http://www.mandrake.tips.4.free.fr/opensuse10.3.review.html
I've found your comments in this thread: http://linux.derkeiler.com/Mailing-Lists/SuSE/2007-11/msg02200.html If those are not your comments, please forgive me; it's not easy to retrieve the proper email address... but please do let me know if I messed up!
Since the opensuse list mentions over 100 mails per day, I don't feel inclined to subscribe, so I'll just mention the following to you directly, mostly related to the tangle of words in that thread with jdd, who jumps to conclusions several times, even though I thought my review was actually clear enough to the contrary. Guess not. Neither he nor I am native English speakers though. So this kind of misunderstanding can happen.
First: thanks for the positive feedback. I'm quite used to getting flamed after mentioning negative points, and it's more rare to get praise.
Now on to the points I'd like to make:
- the installation was definitely fully fresh. The subsection named
'installation' mentions the word 'installation' about 10 times explicitly, and nowhere is the word update or upgrade used in that section, and where used elsewhere it's clear from the context that it's only related to software packages for the openSUSE 10.3 This explains why I was seriously delighted that openSUSE could actually pick up the users from my other systems, including passwords! and also the systems in the bootloader. If it were an upgrade, I would have expected no less. BTW it seems the user data and boot info is taken from the partition on which one installs (from the comments in the above thread), which held a Mandriva 2007.0 installation before with LILO. Colour me even more impressed that openSUSE can manage that!
- the problems with booting and rebooting were to do with a faulty grub in the
root partition, and with my ignorance about whether I should relaunch the installation from dvd or start the half installed and unconfigured system. My only point there was that I would have liked to have some more information than just "now it's time to reboot". I do not blame the faulty grub in the root partition on SUSE, and I'm not sure if this is repeatable, and or a problem with the partitioning of my drive. Note that I intended to use chainloading from the GRUB in the MBR, which I also use to start 2 of the other systems - the GRUB in the mbr is the one from the latest Mandriva 2008.0 installation.
- the reason my other system is not mentioned is that from the 3 other
partitions, all are Mandriva: 64 and 32 bit 2007.1 and 64 bit 2008.0. I tried and managed to keep the whole name Mandriva out of the text of the review. It's not a comparative review, it's a review that looks at things in their own right. The comparative review is still coming. Otherwise, I have two swap partitions and a /home partition - which I didn't use in the openSUSE installation, it was truly and completely _fresh_ and standalone. I count on wiping it. And yes, I do count on giving openSUSE 11 a try in about 5 months.
- sax2 clearly used settings for the i810 driver for the setup with the default
intel graphics driver; I've found out this causes problems only with laptop screens, and maybe only those of 1280x800 resolution - so quite a few less systems than one could think when Intel GMA950 / 945GM is mentioned. Essentially, the modeline of sax2 works with i810 and not with intel - where it's actually perfectly fine (at least in my case) to omit any timing and modelines.
- the locking of the system is caused by issues with the TSST drive in my
system, with the older ata driver there's no problem, with the new libata driver there is this issue - also seen on gentoo, ubuntu and other systems. My Mandriva 2008.0 is fine since they stuck with the old driver, because they _knew_ the new one was problematic. My guess: they knew from: gentoo, ubuntu and openSUSE 10.3...
- by the way, the hardware I used and which I bought as 'Novell Certified' is a
transtec branded MSI Megabook s262 Intel Core 2 Duo Intel GMA950/945GM 3945ABG 12" laptop. The main points: Intel Core 2 Duo, gma950 and ipw3945 are mentioned. I didn't know that the drive was a TSST at all at that point - never cared to check, only came upon this due to the replies in the discussion thread of my article (on some suse forums at http://forums.suselinuxsupport.de/index.php?showtopic=62988) (My hardware info is on my site as well, forgot to link it from the opensuse review page. Sorry.)
- Interesting that a lot of assuming seems to be preferable to some to the
simple questions one could ask to get cleared up - either via that forum link for discussion or via email - via the contact link on my site. Such as: did you do a clean install, did you mess around with x, y, z, etc. It's not like I wouldn't answer those...
- the one problem I 'caused' which can clearly be considered uncommon is the
installation in the root partition to be chainloaded from the grub in the mbr. The following problem are, I believe, totally repeatable:
- intel driver not working with sax2 modelines - unusable graphics
- lockups, admittedly due to faulty firmware in TSST optical drives _BUT_ this
was never seen before because the problem only raises it's head with the new libata driver
- video playback problems
- i586 package installed instead of a x86-64 package using the one-click
installer
- I did have a look at filing a bug for the graphics problems, but I found
enough that were matches for my issues that I didn't. Same for the problem with the lockups and video playback.
Feel free to post this if it's useful and appropriate,
with kind regard, Rob Teng a.k.a. aRTee
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
The Tuesday 2007-11-20 at 20:01 +0100, jdd wrote:
I *never* make an upgrade, and this with any distro (I tried many, specially mandrake or debian for years) because there are always a lot of problems.
And I *always* make an upgrade. This machine has been upgraded from 7.3 up to 10.3 with only one interruption in the flowchain.
As a matter of fact, the 10.2 --> 10.3 upgrade was the smoother of them all, and the 7.3 --> 8.1 the worst.
But the reviewer did a fresh install - as any reviewer should do.
- -- Cheers, Carlos E. R.
Richard Creighton wrote:
jdd wrote:
Richard Creighton wrote:
eminently fair and for the most part, accurate. I think any inaccuracies could be attributable to unfamiliarity rather than bias. If this reviewer hadn't been what I would call a 'Power-user', and had instead been someone converting from Windows, not only would they probably not have had a successful install
completely wrong.
because he is a power user, he have four linux on his laptop. It's already good if he could install at all. He don't say what answers he give to the questions, but, for example, is glad to have Yast asking if he wants to keep old users. Do you think he could have such question on a fresh install? I beg no. He is making an update. and what was the old Linux? who knows?
Balderdash! Are you saying that because he is experienced and that he has other distros or OS's on his system that he can't write an objective review and if he wants to upgrade his system from an older version instead of doing a 'fresh install', does that make his review less valuable. Few people I know want to erase months or years of configuration effort by the equivilent of FORMAT C: just to get a 'fresh install'. I have communicated with many from around the world that simply wanted to upgrade from 10.2 to 10.3 and the system failed miserably. Why should that not be reported if it is a fact? To insist that the only VALID test of an operating system or distro like 10.3 is to wipe your disk and start fresh in order to avoid problems is ludercrous and utterly unrealistic. It is lucky that he was a 'power user' or the review would likely have been a LOT less to your obviously biased liking.
This group was asked earlier by management of openSuSE.org to try out other distros and report back our findings. I can't think of a better report than a user of another distro giving a reasonably fair evaluation of *our* product. If it has warts, we should be willing to admit it and do something constructive about it rather than burying our collective heads in the sand and saying '...but he is a power user on a special machine' and ignoring what he had to say.
so may be he got wireless working because it was working on the old install and so the firmware was at hand (how could he have it if not?) and x don't working because the old install was not completely compatible with the old one.
He did give credit where credit was due. He only said that in many cases, Yast, is not very intuitive and that also is a fact. It is a dammed good program, but that doesn't mean it is perfect and can't stand improvement.
any intelligent review must be done on a fresh (not new) computer if it wants to be other than individual experience
Balderdash! Again! I think doing it on a so-called 'fresh' machine is very Microsoft in its' mentality. Who did you say you work for? Real life dictates that most installs will be upgrades not fresh installs and tests that ignore that fact are themselves deficient. So, if as you say, he did an upgrade, (which I didn't read from his report but might have missed), that makes the test even more valid, not less.
this is the install of a power user upgrading an old install.
So, why does SuSE give the option to UPGRADE....if it isn't a valid method of getting the distro onto ones' disk, then it should simply erase everything and start fresh. That should make you happy and you can write your own review. Be sure and complain about the loss of all your favorite programs and settings and the amount of time it takes to relocate the programs and set them up again from scratch. It is your
This is why smart people make /home, /opt, and /local to be separate filesystems (or at least make /local a symbolic link to something like /home/local and/or /opt to be a symbolic link to /home/opt.) This should be the DEFAULT set-up for a new installation, because the uninitiated has absolutely no idea that having /home, /local and /opt on the root filesystem is setting them up for major headaches when they want to install the system.
In fact, in the install, it would be REALLY super-nice if I could specify the creation of any symbolic links and target directories before ANY packages are installed.
machine, after all, but for me, I want to know if I can still have success upgrading or if I can have dual or quadruple boot to other distros or even Windoze. Any review that leaves out that info is less valuable IMO. And lest we forget, right up front, he said that it had been a long time since he had used SuSE and he felt is was time to give it another try....which I think he did. So, again, SuSE management *should* read his review because it is not written from one of the lemmings, but from someone actually spending a bit of time and trouble evaluating the competition.
Richard
jdd
On Tue, 2007-11-20 at 23:36 -0500, Aaron Kulkis wrote:
This is why smart people make /home, /opt, and /local to be separate filesystems (or at least make /local a symbolic link to something like /home/local and/or /opt to be a symbolic link to /home/opt.) This should be the DEFAULT set-up for a new installation, because the uninitiated has absolutely no idea that having /home, /local and /opt on the root filesystem is setting them up for major headaches when they want to install the system.
In fact, in the install, it would be REALLY super-nice if I could specify the creation of any symbolic links and target directories before ANY packages are installed.
While I certainly agree with that smart approach to filesystems, at the very least /home. But how would such a default setup work from installation? I don't think any distro is possibly intuitive enough to know whether you wanted it in a primary partition, extended partition, or LVM. And that's just assuming you have one drive on your machine.
Partitioning strategies are still pretty much left to the knowledgeable folks who understand partitioning rules and strategies. And probably should be. How many end-users understand those concepts? And you DO have to understand those concepts in order to make the best decision. There's no way Novell or anyone else can predict a) # of drives you have, b) what partitions currently exist and what they're used for and c) what type of partitioning strategy you intend to use on your computer.
Bryen wrote:
On Tue, 2007-11-20 at 23:36 -0500, Aaron Kulkis wrote:
This is why smart people make /home, /opt, and /local to be separate filesystems (or at least make /local a symbolic link to something like /home/local and/or /opt to be a symbolic link to /home/opt.) This should be the DEFAULT set-up for a new installation, because the uninitiated has absolutely no idea that having /home, /local and /opt on the root filesystem is setting them up for major headaches when they want to install the system.
In fact, in the install, it would be REALLY super-nice if I could specify the creation of any symbolic links and target directories before ANY packages are installed.
While I certainly agree with that smart approach to filesystems, at the very least /home. But how would such a default setup work from installation? I don't think any distro is possibly intuitive enough to know whether you wanted it in a primary partition, extended partition, or LVM. And that's just assuming you have one drive on your machine.
In any event, the default (on every installation, not just SuSE) of making just two partitions: a huge root partition, and a swap partition at the end of free disk space, is just insane.
Partitioning strategies are still pretty much left to the knowledgeable folks who understand partitioning rules and strategies. And probably should be. How many end-users understand those concepts? And you DO have to understand those concepts in order to make the best decision. There's no way Novell or anyone else can predict a) # of drives you have,
every SuSE install I've done, going all the way back to the 6.x days, has detected all of my hard drives.
b) what partitions currently exist and what they're used for and
Every SuSE install I've done has detected currently existing partitions.
c) what type of partitioning strategy you intend to use on your computer.
At the very least:
root-partition ; swap partition ; /home partition
Partition sizes for filesystems don't need to be calculated until AFTER the user selects what software they want (unless the person doing the install selects for it). Filesystem usage from selected packages should be easy to calculate -- the installer is already calculating how much space will be used on each filesystem when selecting software after specifying a custom partition scheme(*) -- build some rules for various parts of the hierarchy based on a minimum suggested "cushion" of unused space in both megabytes (say for the spool directories in /var), AND in terms of percentages (say, %20 percent each for /lib, /usr/lib, and /usr/X11/lib to allow for growth).
(*) I know this, because as you do so, there are little "meters" in the install screen that show you how much space is going to be used in each hierarchy -- of course, adding install-time symbolic links would complicate this somewhat.
A modern SuSE installation, even an "everything and the kitchen sink" installation is only a small fraction (1/5 or less) of most new disk drives. In the even of detecting a large hard drive, the install system should immediately suggest a large number of partitions (/var, /usr, /home, /opt, /tmp, and /local, and /srv (which REALLY should be a symbolic link to something like /local/srv), so as to keep filesystem write() activity on the root filesystem to the essential minimum.
Remember, every write to a filesystem is an opportunity for corruption to occur (yes, even in journalled filesystems, don't ask me about my experiences with ReiserFS)...and if the space is detected and available, the default behavior of the partitioner should be making seperate partitions for those hierarchies with a lot of write()-activity ( /var, /tmp, /home, /srv), and making the suggestion (or maybe even defaulting also) that /local and /opt should also be on separate filesystems if it is anticipated that the OS could be upgraded before the machine reaches the end of its useful life AND the admin doesn't want to back-up+restore / reinstall whatever ends up in those /local and /opt.
I know these are not trivial things to do... and unfortunately, my programming skills are far too rusty (20 years of nonuse) to contribute... but doing this sort of thing would significantly improve the installation on a new machine, especially for newbies who, quite frankly, don't know any better (the default installation from both M$ and Windows OEMs is quite bad. Windows should have as a minimum three partitions: C: for the OS, and two more -- one for application programs, and one for data (including user's desktops)...but M$ default behavior is to re-assign drive letters if a new hard disk is added, so that the first partition on the new disk is inserted BEFORE the 2nd, 3rd, 4th.. partitions that are on the same physical drive as C:).
Aaron Kulkis wrote:
This should be the DEFAULT set-up for a new installation,
but it's *exactly* what YaST did when I installed 10.3!!
this was on a 160Gb disk with the standard vista default install of having three partitions (one for system backup, one for vista root and one for vista data - the laptop already quoted in one of my post).
this data partition is really wonderfull to install linux on a dual boot computer without fips problems.
on my computer it was 70Gb and Yast used 20Gb for root and 40 Gb for /home (not to mention 2Gb for swap)
may be your disk was found too small to do so.
In fact, in the install, it would be REALLY super-nice if I could specify the creation of any symbolic links and target directories before ANY packages are installed.
? I don't like too many automatic things, mostly at install time. What I should like is a "post-install tweaks" YaST module... At install time I want a really fast install giving a working system (mouse and X)
I want to insist on a thing: relese notes are not complete enough (and most don't even read them :-), so I don't know when was included the "sax2-vesa" link that solve most if not all the screen problems for a basic install)
but the "TAB" completion is a very valuable openSUSE tool. OpenSUSE because openSUSE uses it the most about all distributions.
type sax <TAB> and you get all the sax main options, rc <TAB> and you get all the init options, start <TAB> and you get all the startx/xfce options. this only makes openSUSE invaluable.
and about of popularity, don't forget the most popular distribution now (ubuntu) is
* debian based with the good of Debian but openSUSE friendlyness * send for free as many cd's as one want anywhere in the world...
don't forget neither that today most personal computers are laptops (my LUG room is filled with laptops when meeting time, I had never seen that before) and laptops are always been the most difficult hardware to cope with.
don't forget again than hardware recognition is a kernel problem, so most of the time equally spread among distribution, the newer being the better
jdd
On 11/21/2007 jdd wrote:
and about of popularity, don't forget the most popular distribution now (ubuntu) is
- debian based with the good of Debian but openSUSE friendlyness
- send for free as many cd's as one want anywhere in the world...
The only gripe I have with Ubuntu is a tool like Yast, and Sax. Those two would be high on my wish list.
The only gripe I have with Ubuntu is a tool like Yast, and Sax. Those two would be high on my wish list.
They are working on that.... but their solution is nowhere near as good or as slick as YAST. Ubuntu also does some odd things with where some things end up. Like the web server ends up placing everything in /var/www instead of /srv/www (maybe that is just my familiarity with SUSE's oddities there though).
When people ask me, I usually say that if they want a simple desktop, Ubuntu works fine. If they want to run a MySQL database, run a webserver etc etc in addition to the desktop, then openSUSE wins by a few light years despite it's warty bits. :-)
The article that started this thread makes a lot of really good points. While a lot of the problems the author got himself into were a product of his own tinkering, those problems are very real. I see a lot of the same kinds of dufficulties when I am helping newbie users install their very first openSUSE. I don't run into those problems (or at least am not so bothered by them) because I have been using SUSE since 6.0. I'm used to the warts. I do hope that the comments made in the article are taken to heart and at least some of the problems attended to in the next release or two. Some of those problems have been around since forever.
C.
Dňa Wednesday 21 November 2007 05:36:35 Aaron Kulkis ste napísal:
This is why smart people make /home, /opt, and /local to be separate filesystems (or at least make /local a symbolic link to something like /home/local and/or /opt to be a symbolic link to /home/opt.) This should be the DEFAULT set-up for a new installation, because the uninitiated has absolutely no idea that having /home, /local and /opt on the root filesystem is setting them up for major headaches when they want to install the system.
AFAIK /home is by default on a different partition. The other ones are rather special for people installing a lots of additional software.
In fact, in the install, it would be REALLY super-nice if I could specify the creation of any symbolic links and target directories before ANY packages are installed.
This sounds like super-expert stuff. You can do it with add-on product, but I don't think this will be ever implemented.
Stano
Stanislav Visnovsky wrote:
Dňa Wednesday 21 November 2007 05:36:35 Aaron Kulkis ste napísal:
This is why smart people make /home, /opt, and /local to be separate filesystems (or at least make /local a symbolic link to something like /home/local and/or /opt to be a symbolic link to /home/opt.) This should be the DEFAULT set-up for a new installation, because the uninitiated has absolutely no idea that having /home, /local and /opt on the root filesystem is setting them up for major headaches when they want to install the system.
AFAIK /home is by default on a different partition. The other ones are rather special for people installing a lots of additional software.
In fact, in the install, it would be REALLY super-nice if I could specify the creation of any symbolic links and target directories before ANY packages are installed.
This sounds like super-expert stuff. You can do it with add-on product, but I don't think this will be ever implemented.
No, I doubt it ever will be... it's a "wish list" item.
BTW, what "add-on product" is going to do this in the middle of installation?
For example, I might not want /var/fonts on the /var partition... but would instead, prefer to put it in some place like /opt/fonts or /local/fonts )
Right now, it involves doing the install, then going into single user mode, moving the fonts directory and creating the symbolic link from /var/fonts to /local/fonts...
Not much of a problem on all IDE systems, but when you're dealing with SCSI disks, partition size can still be a factor.
Dňa Wednesday 21 November 2007 11:17:02 Aaron Kulkis ste napísal:
Stanislav Visnovsky wrote:
Dňa Wednesday 21 November 2007 05:36:35 Aaron Kulkis ste napísal:
This is why smart people make /home, /opt, and /local to be separate filesystems (or at least make /local a symbolic link to something like /home/local and/or /opt to be a symbolic link to /home/opt.) This should be the DEFAULT set-up for a new installation, because the uninitiated has absolutely no idea that having /home, /local and /opt on the root filesystem is setting them up for major headaches when they want to install the system.
AFAIK /home is by default on a different partition. The other ones are rather special for people installing a lots of additional software.
In fact, in the install, it would be REALLY super-nice if I could specify the creation of any symbolic links and target directories before ANY packages are installed.
This sounds like super-expert stuff. You can do it with add-on product, but I don't think this will be ever implemented.
No, I doubt it ever will be... it's a "wish list" item.
BTW, what "add-on product" is going to do this in the middle of installation?
For example, I might not want /var/fonts on the /var partition... but would instead, prefer to put it in some place like /opt/fonts or /local/fonts )
Right now, it involves doing the install, then going into single user mode, moving the fonts directory and creating the symbolic link from /var/fonts to /local/fonts...
Not much of a problem on all IDE systems, but when you're dealing with SCSI disks, partition size can still be a factor.
What you can do is to add additional steps during the installation to do whatever you need to do.
You can do it easily for autoyast, for normal install,you need to create an add-on product with a workflow.
Stano
On Tue November 20 2007 10:12, jdd wrote:
this is the install of a power user upgrading an old install.
Sorry, but your assumption is wrong. openSuse 10.3 now recognizes an existing installation and retrieves info from the old /etc/passwd before formatting the hard drive, even when a complete fresh installation is performed.
I know, because I never upgrade. I prefer to do a fresh install of my root partition, keeping only the home partition intact.
Carlos F Lange wrote:
On Tue November 20 2007 10:12, jdd wrote:
this is the install of a power user upgrading an old install.
Sorry, but your assumption is wrong. openSuse 10.3 now recognizes an existing installation and retrieves info from the old /etc/passwd before formatting the hard drive, even when a complete fresh installation is performed.
I know, because I never upgrade. I prefer to do a fresh install of my root partition, keeping only the home partition intact.
I also keep /opt and /local around, so that I don't have to re-install other stuff that I've downloaded and installed.
Dňa Tuesday 20 November 2007 18:12:01 jdd ste napísal:
Richard Creighton wrote:
eminently fair and for the most part, accurate. I think any inaccuracies could be attributable to unfamiliarity rather than bias. If this reviewer hadn't been what I would call a 'Power-user', and had instead been someone converting from Windows, not only would they probably not have had a successful install
completely wrong.
because he is a power user, he have four linux on his laptop. It's already good if he could install at all. He don't say what answers he give to the questions, but, for example, is glad to have Yast asking if he wants to keep old users. Do you think he could have such question on a fresh install? I beg no. He is making an update. and what was the old Linux? who knows?
The installer is able to detect if there are old installations and get the list of the users also when doing a clean install side-by-side with other linux installation. I believe he referred to this functionality.
Stano
Stanislav Visnovsky wrote:
The installer is able to detect if there are old installations and get the list of the users also when doing a clean install side-by-side with other linux installation. I believe he referred to this functionality.
I never see this when doing a fresh install (even with multi distro), may be I missed it
jdd
On 11/21/2007 03:55 PM, Stanislav Visnovsky wrote:
The installer is able to detect if there are old installations and get the list of the users also when doing a clean install side-by-side with other linux installation. I believe he referred to this functionality.
I pleasantly discovered that when I did an update from 10.2. I decided when I had 9.3, and upgraded to 10.2, to add some disks and install on a fresh root. This time, I installed 10.3 cleanly on the old and reformatted former 9.3 / (home is on its own). It obviously scanned the 10.2 root and copied over all the users, saving me a ton of work. Next time I will just upgrade 11 over the 10.2 root. I remember how long it took to re-enter everyone when I upgraded to 8.0. Well done!
Joe Morris (NTM) wrote:
On 11/21/2007 03:55 PM, Stanislav Visnovsky wrote:
The installer is able to detect if there are old installations and get the list of the users also when doing a clean install side-by-side with other linux installation. I believe he referred to this functionality.
I pleasantly discovered that when I did an update from 10.2. I decided when I had 9.3, and upgraded to 10.2, to add some disks and install on a fresh root. This time, I installed 10.3 cleanly on the old and reformatted former 9.3 / (home is on its own). It obviously scanned the 10.2 root and copied over all the users, saving me a ton of work. Next time I will just upgrade 11 over the 10.2 root. I remember how long it took to re-enter everyone when I upgraded to 8.0. Well done!
it's enough to copy /etc/passwd, groups and shadow
jdd
Is there some reason you're sending out old mail? I've seen several old posts from you today.
jdd wrote:
Joe Morris (NTM) wrote:
On 11/21/2007 03:55 PM, Stanislav Visnovsky wrote:
The installer is able to detect if there are old installations and get the list of the users also when doing a clean install side-by-side with other linux installation. I believe he referred to this functionality.
I pleasantly discovered that when I did an update from 10.2. I decided when I had 9.3, and upgraded to 10.2, to add some disks and install on a fresh root. This time, I installed 10.3 cleanly on the old and reformatted former 9.3 / (home is on its own). It obviously scanned the 10.2 root and copied over all the users, saving me a ton of work. Next time I will just upgrade 11 over the 10.2 root. I remember how long it took to re-enter everyone when I upgraded to 8.0. Well done!
it's enough to copy /etc/passwd, groups and shadow
jdd
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Richard Creighton wrote:
Sid Boyce wrote:
<snip>
I didn't have problems as grave as the reviewer. I'm not happy with package management, but frankly I think this is just an excuse to pile on the team; 10.3 does what I need it to. But that's just my opinion.
Moving on, I see no point in cross-posting this. This only serves to fragment the input of people who might have something to say on this thread. I recommend the OP either kill the thread on this list or on opensuse-project and wherever else it might have been cross-posted.
- -- http://www.DonAssad.com jabber ID: josef.assad@gmail.com
Please consider the environment; do you really need to print out this e-mail?
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Josef Assad wrote:
Richard Creighton wrote:
Sid Boyce wrote:
<snip>
I didn't have problems as grave as the reviewer. I'm not happy with package management, but frankly I think this is just an excuse to pile on the team; 10.3 does what I need it to. But that's just my opinion.
Moving on, I see no point in cross-posting this. This only serves to fragment the input of people who might have something to say on this thread. I recommend the OP either kill the thread on this list or on opensuse-project and wherever else it might have been cross-posted.
s/opensuse-project/opensuse-factory/
:)
JA
- -- http://www.DonAssad.com jabber ID: josef.assad@gmail.com
Please consider the environment; do you really need to print out this e-mail?
Richard Creighton wrote:
Sid Boyce wrote:
http://www.mandrake.tips.4.free.fr/opensuse10.3.review.html Mentions of "for the uninitiated" sounds about right. To sidestep any problems with the DVD, I've installed from Factory. Looks like the DVD should have been delayed until it was compared with Factory for stability. I think the reviewer attempted to be fair (he points to the URL to show where he is coming from). Perhaps some pertinent pointers for 11.0. Regards Sid.
http://www.mandrake.tips.4.free.fr/opensuse10.3.review.html should be MUST READING for management at Novell and at openSuSE.org. It is eminently fair and for the most part, accurate. I think any inaccuracies could be attributable to unfamiliarity rather than bias. If this reviewer hadn't been what I would call a 'Power-user', and had instead been someone converting from Windows, not only would they probably not have had a successful install, but would have an article that Microsoft would probably have printed in full page ads throughout the world. As it is, it should wake up Senior Management in SuSE.
regards Richard
Richard,
You hit the nail on the head. Good job! Just think about how the review would have been different if Novell had taken the time to QA and fix just the issues listed in that article. (pause and think about it)
All of the frustration and waste of time experienced and expended by each of us would have been spared and the support headaches for Novell would have been eliminated. I am committed to openSuSE after having come from Mandrake when that distribution imploded and went through its "dark times" (8.0 - 10.1). What caused the mass exodus from Mandrake?
The exodus was caused by the corpratizing of what was a fantastic disro through 7.2. Quality developers committed to thorough QA, putting out a reliable product when it was ready. Once Mandrake went public, the developers were stripped of control over the distribution. The business types made business decisions to adhere to an insane release cycle and QA and reliability went down the tubes. It has taken Mandrake several *years* to recover, but they are getting there. A costly few years it took to find the correct balance between the business and software sides of the house.
In business, as in virtually every facet of life, history can be a great teacher for all intelligent enough to listen and learn. The problems encountered in 10.3 pale by comparison to the Mandrake debacle, but they indicate a troubling change in direction. When I moved to SuSE at (8.0, 8.2?) the distro was the best on the market - hands down. That, even with the yast ftp install that required booting from a floppy. It worked, it worked every time, and gotchas were a very rare occurrence.
arTee's article is extremely well written and a candid, honest and fair account of the 10.3 release. I applaud the time, the dedication and the effort he expended to give back to the OSS community and provide an opportunity for openSuSE to access how the distribution was released and learn from a bit of history.
Don't get me wrong, I'm not knocking openSuSE at all. I still believe it is the best distribution on the planet. But experiencing the exact same install problems described in the article and after spending the past 50 days sorting out lockups and many of the other problems described, it has left me shaking my head at the sheer amount of time required to get 10.3 fleshed out. arTee's point is well made, and I agree. A vast majority of those that aren't already sufficiently Linux literate trying 10.3 as a first Linux try, would have thrown in the towel right after the install glitch and would never have looked back or gotten to experience the elegance of this distro.
I know the developers are doing a great job, but I truly wonder if some of the critical decisions are being made by the corporate folks ignoring the better judgment of the developers. The article provides great feedback and a great opportunity to learn. Lest another article with the same candid, honest and fair account is to follow -- after the release of 11.0.
Richard Creighton wrote:
Sid Boyce wrote:
http://www.mandrake.tips.4.free.fr/opensuse10.3.review.html Mentions of "for the uninitiated" sounds about right. To sidestep any problems with the DVD, I've installed from Factory. Looks like the DVD should have been delayed until it was compared with Factory for stability. I think the reviewer attempted to be fair (he points to the URL to show where he is coming from). Perhaps some pertinent pointers for 11.0. Regards Sid.
http://www.mandrake.tips.4.free.fr/opensuse10.3.review.html should be MUST READING for management at Novell and at openSuSE.org. It is eminently fair and for the most part, accurate. I think any inaccuracies could be attributable to unfamiliarity rather than bias. If this reviewer hadn't been what I would call a 'Power-user', and had instead been someone converting from Windows, not only would they probably not have had a successful install, but would have an article that Microsoft would probably have printed in full page ads throughout the world. As it is, it should wake up Senior Management in SuSE.
Some of the problems, like the X resolution mis-match, are just plain unacceptable. I immediately became leery of RedHat when they went to their current business-model: "We're going to focus our support on 'Enterprise' installations, which we will sell at over $1000/DVD, and a "desktop" distro to connect to them....and then the distro that MOST of you want ... that is, light-server capabilities (a webserver app or two, a couple database engines, etc in the distro) -- we're going to just punt support for that to the userbase and call it 'Fedora'"
Not only did I quit buying from RedHat -- I quit USING RedHat, too. I see that Novell's decision to imitate RedHat is causing my worst fears to be realized with the SuSE distro -- the one distro that I want...which I'm willing to pay around $100 US for... is not getting supported properly, and it's sounding like, even for the cost of burning a download, the value-proposition is somewhat iffy.
This is totally sad. I've been a die-hard SuSE fan since the 6.x days, and I've forked over probably $400 over the years for the various incarnations which are now the "opensuse" distribution.
And to tell you the truth, I would much rather CONTINUE to pay $80 to $100 every year for a DVD which I know is well-supported and has basic sanity checking done by someone who is GETTING PAID TO MAKE SURE THE BASICS ARE RIGHT than to download something that was only checked in a half-assed sort of way because NOBODY is geting paid to make sure that even the basic installation process is doing EVERYTHING that it's supposed to do -- such as making sure that the X-server resolution matches the desktop resolution.
There's a certain point where volounteerism doesn't quite get the job done because dotting the i's and crossing the t's on certain parts just isn't as "interesting" and exciting as adding more features.
If SuSE wants to have a "loss leader" distribution, it should be a 'desktop' distro (which used to sell for $20 to $40 less than the "professional" distro), not the one which an overwhelming percentage of SuSE users are both able AND willing to fork over cold hard cash.
Aaron Kulkis wrote:
Not only did I quit buying from RedHat -- I quit USING RedHat, too. I see that Novell's decision to imitate RedHat is causing my worst fears to be realized with the SuSE distro -- the one distro that I want...which I'm willing to pay around $100 US for... is not getting supported properly, and it's sounding like, even for the cost of burning a download, the value-proposition is somewhat iffy.
This is totally sad. I've been a die-hard SuSE fan since the 6.x days, and I've forked over probably $400 over the years for the various incarnations which are now the "opensuse" distribution.
Reviewing the boxes and manuals laying around the room, it looks like I spent about US $600 on "Suse Professional" distributions since the 6.x days.
Aaron Kulkis wrote:
Aaron Kulkis wrote:
Not only did I quit buying from RedHat -- I quit USING RedHat, too. I see that Novell's decision to imitate RedHat is causing my worst fears to be realized with the SuSE distro -- the one distro that I want...which I'm willing to pay around $100 US for... is not getting supported properly, and it's sounding like, even for the cost of burning a download, the value-proposition is somewhat iffy.
This is totally sad. I've been a die-hard SuSE fan since the 6.x days, and I've forked over probably $400 over the years for the various incarnations which are now the "opensuse" distribution.
Reviewing the boxes and manuals laying around the room, it looks like I spent about US $600 on "Suse Professional" distributions since the 6.x days.
That is a lot of money.
Otoh, imagine you started to use Windows in the Suse 6.x days. How much would you have spent on this?
If you review those two figures, it gets clear, that opensuse cannot be the foundation for the Novell Business while windows licenses are (part of) the foundation for the MS business-
So as sad as it is, we do not pay (sufficiently) for the software that we use. Therefore, our influence on it's development is very limited, financewise.
Kind regards Eberhard
Eberhard Roloff wrote:
Aaron Kulkis wrote:
Aaron Kulkis wrote:
Not only did I quit buying from RedHat -- I quit USING RedHat, too. I see that Novell's decision to imitate RedHat is causing my worst fears to be realized with the SuSE distro -- the one distro that I want...which I'm willing to pay around $100 US for... is not getting supported properly, and it's sounding like, even for the cost of burning a download, the value-proposition is somewhat iffy.
This is totally sad. I've been a die-hard SuSE fan since the 6.x days, and I've forked over probably $400 over the years for the various incarnations which are now the "opensuse" distribution.
Reviewing the boxes and manuals laying around the room, it looks like I spent about US $600 on "Suse Professional" distributions since the 6.x days.
That is a lot of money.
Not really
Otoh, imagine you started to use Windows in the Suse 6.x days. How much would you have spent on this?
For the same amount of software? $500,000 easily.
Database engines that don't fall over just because 50 people are accessing the database are quite expensive in MS land.
If you review those two figures, it gets clear, that opensuse cannot be the foundation for the Novell Business while windows licenses are (part of) the foundation for the MS business-
Look at the figures above.
So as sad as it is, we do not pay (sufficiently) for the software that we use. Therefore, our influence on it's development is very limited, financewise.
I'm more than happy to pay $$$ to SuSE to put out a good distro.
Kind regards Eberhard
On Wed, Nov 21, 2007 at 09:50:25AM +0100, Eberhard Roloff wrote:
Aaron Kulkis wrote:
Aaron Kulkis wrote:
Not only did I quit buying from RedHat -- I quit USING RedHat, too. I see that Novell's decision to imitate RedHat is causing my worst fears to be realized with the SuSE distro -- the one distro that I want...which I'm willing to pay around $100 US for... is not getting supported properly, and it's sounding like, even for the cost of burning a download, the value-proposition is somewhat iffy.
This is totally sad. I've been a die-hard SuSE fan since the 6.x days, and I've forked over probably $400 over the years for the various incarnations which are now the "opensuse" distribution.
Reviewing the boxes and manuals laying around the room, it looks like I spent about US $600 on "Suse Professional" distributions since the 6.x days.
That is a lot of money.
Otoh, imagine you started to use Windows in the Suse 6.x days. How much would you have spent on this?
If you review those two figures, it gets clear, that opensuse cannot be the foundation for the Novell Business while windows licenses are (part of) the foundation for the MS business-
So as sad as it is, we do not pay (sufficiently) for the software that we use. Therefore, our influence on it's development is very limited, financewise.
No, not at all, for SuSE didn't pay (sufficiently) for the software that they sell :)
And, your influence on its development is far greater than any closed operating system, as you can talk directly to the developers, and actually contribute code and features to it if you want to.
thanks,
greg k-h
-
Aaron Kulkis -
Billie Walsh -
Bryen -
Carlos E. R. -
Carlos F Lange -
Clayton -
David C. Rankin -
Eberhard Roloff -
Greg KH -
James Knott -
jdd -
Joe Morris (NTM) -
Josef Assad -
Richard Creighton -
Richard Creighton -
Stanislav Visnovsky