Whilst I'm looking forward to RC2 ...
I can't help noticing a bit of a problem. With a very aggressive release-schedule such as "every 2 weeks", chances of genuine bugs being 1) reported, 2) diagnosed and 3) fixed in that timeframe are very minimal. When I say "genuine" I discount typos, apparmor bureacrazy, and other minors that can be dealt with by level 1 support. I've got a couple of open reports (159731, 136742, 159727, 153585), and whilst an aggressive release-schedule is probably a Good Thing(R), it puts a lot of unnecessary strain on testers. For bugs/problems that aren't positively fixed before the next release, it's usually "please install <next release> and see if problem has gone away". Perhaps I'm just whining or venting some steam, but the aggressive release-schedule isn't doing me much good. /Per Jessen, Zürich
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
I can't help noticing a bit of a problem. With a very aggressive release-schedule such as "every 2 weeks", chances of genuine bugs being
1) reported, 2) diagnosed and 3) fixed
in that timeframe are very minimal.
This is my greatest fear.
When I say "genuine" I discount typos, apparmor bureacrazy, and other minors that can be dealt with by level 1 support. I've got a couple of open reports (159731, 136742, 159727, 153585), and whilst an aggressive release-schedule is probably a Good Thing(R), it puts a lot of unnecessary strain on testers. For bugs/problems that aren't positively fixed before the next release, it's usually "please install <next release> and see if problem has gone away".
Perhaps I'm just whining or venting some steam, but the aggressive release-schedule isn't doing me much good.
I too am very concerned about this and agree 1000%. I wonder the same
thing. I do not want to complain but I have real concerns.
- --
Boyd Gerber
On Thursday 20 April 2006 16:10, Boyd Lynn Gerber wrote:
I can't help noticing a bit of a problem. With a very aggressive release-schedule such as "every 2 weeks", chances of genuine bugs being
1) reported, 2) diagnosed and 3) fixed
in that timeframe are very minimal.
This is my greatest fear.
When I say "genuine" I discount typos, apparmor bureacrazy, and other minors that can be dealt with by level 1 support. I've got a couple of open reports (159731, 136742, 159727, 153585), and whilst an aggressive release-schedule is probably a Good Thing(R), it puts a lot of unnecessary strain on testers. For bugs/problems that aren't positively fixed before the next release, it's usually "please install <next release> and see if problem has gone away".
Perhaps I'm just whining or venting some steam, but the aggressive release-schedule isn't doing me much good.
I too am very concerned about this and agree 1000%. I wonder the same thing. I do not want to complain but I have real concerns.
I think they are trying to implement XP, you know shorter cycle, more release and also more feedback from the customer (us). And iirc my XP book, if you find some major problem you adapt the schedule ... and I believe they are doing this too so for me it's not a concern, just another way of producing high quality software products. Mathieu -- Hollywood is where if you don't have happiness you send out for it. -- Rex Reed
On Thursday 20 April 2006 22:10, Boyd Lynn Gerber wrote:
Perhaps I'm just whining or venting some steam, but the aggressive release-schedule isn't doing me much good.
I too am very concerned about this and agree 1000%. I wonder the same thing. I do not want to complain but I have real concerns.
It's not long ago we went from weekly to bi-weekly releases ;) I think that was a correct decision though. However I don't think there are many lessons that can be learned from the 10.1 development process. Other than too much new immature stuff has been put in, if SUSE Linux is supposed to be a nice stable distro in it's own right - and not just a SLED test platform. However I trust that the problems we're experiencing now won't happen again next time around - 10.1 is an extraordinary release. I expect 10.2 won't be nearly as hard on testers and devs as 10.1 has been/is. I hope I'm not kidding myself ;) cb400f
On Fri, Apr 21, 2006 at 12:36:01AM +0200, Martin Schlander wrote:
Other than too much new immature stuff has been put in, if SUSE Linux is supposed to be a nice stable distro in it's own right - and not just a SLED test platform.
I don't think it is 'just' a SLED test platform, I think it is among other things also a SLED test platform.
However I trust that the problems we're experiencing now won't happen again next time around - 10.1 is an extraordinary release. I expect 10.2 won't be nearly as hard on testers and devs as 10.1 has been/is.
I hope I'm not kidding myself ;)
I am sure the developers will miss the presure and think of something else to do in the last 2 months. ;-D houghi -- Nutze die Zeit. Sie ist das Kostbarste, was wir haben, denn es ist unwiederbringliche Lebenszeit. Leben ist aber mehr als Werk und Arbeit, und das Sein wichtiger als das Tun - Johannes Müller-Elmau
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On Thursday 20 April 2006 22:10, Boyd Lynn Gerber wrote:
Perhaps I'm just whining or venting some steam, but the aggressive release-schedule isn't doing me much good.
I too am very concerned about this and agree 1000%. I wonder the same thing. I do not want to complain but I have real concerns.
It's not long ago we went from weekly to bi-weekly releases ;)
I think that was a correct decision though.
I understand the change. My problem is with all the last minute package ... (installation/support/upgrade/method-changes...) I feel like too much could go untested or HW/BIOS/System problems not in test machines. Once it is released I think it may be used in untested ways with problem not seen in all our quick checking. A lot of the problems are usually caught in the alpha and early betas. The package manangement at such a late time frame in release cycle leaves me really uneasy. I still do not get clean installations to the level of previous releases/RC canadits with 10.1.
However I don't think there are many lessons that can be learned from the 10.1 development process. Other than too much new immature stuff has been put in, if SUSE Linux is supposed to be a nice stable distro in it's own right - and not just a SLED test platform. However I trust that the problems we're experiencing now won't happen again next time around - 10.1 is an extraordinary release. I expect 10.2 won't be nearly as hard on testers and devs as 10.1 has been/is.
At such a late time frame... Usually I have many clean installs with the RC's. I have not had one. Also, I am concerned with the HW support or possible missing support that was in many previous releases. I am not sure how the NO non-GPL HW drivers will finally be. I am still struggling testing Supported HW that has bugs or issues. They all are in bugzilla that I have looked up. Some I can not get to the point of understanding what is going wrong. Things that we normally do not fight at this stage are causing a lot of work. I fear there may be subtle things that no one has got to because of the current bugs.
I hope I'm not kidding myself ;)
I too. I think the last minute changes caused this. I do not think we
will see this done again because of what has happened. I think it is the
biggest lesson learned. I have most of my machines with only 128 M or 256
M memory. Current releases are more demanding on memory. I usually have
1G - 2G swap space and they are experiencing a lot of paging/usage. I
only have one test machine with more than 256 M memory. I do not want to
sacrifice a production machine with the current state of RC's. I usually
use one production machine at this stage, but I just can not afford to
have it down and not working.
Hence my concerns.
Thanks for such a good distribution.
- --
Boyd Gerber
On Thu, Apr 20, 2006 at 07:06:04PM -0600, Boyd Lynn Gerber wrote:
The package manangement at such a late time frame in release cycle leaves me really uneasy.
Yes, and it has also been explained why that was done. Dragging it on and on does nobody any good. Nobody likes it. The decision has been taken, now let us move on. I am looking at it in a different way. If Novell would not have opend up the whole process, then the same decisions would have been taken. The difference (apart from many more bugs being reported and repaired) is that now you suddenly see how things work, while at other times you would just sit back and wait what would be happening. <snip>
At such a late time frame... Usually I have many clean installs with the RC's.
Appart from 10.0, what beta versions of SUSE have you tested? I have seen none. Could be that they were even worse and completely uninstallable untill the last day. houghi -- Nutze die Zeit. Sie ist das Kostbarste, was wir haben, denn es ist unwiederbringliche Lebenszeit. Leben ist aber mehr als Werk und Arbeit, und das Sein wichtiger als das Tun - Johannes Müller-Elmau
houghi wrote:
Appart from 10.0, what beta versions of SUSE have you tested? I have seen none. Could be that they were even worse and completely uninstallable untill the last day.
What about all of 10.1 ? Plenty of betas there. I think I started out with beta3, then 5, 6, 8, 9 and rc1. /Per Jessen, Zürich
On Fri, Apr 21, 2006 at 09:10:28AM +0200, Per Jessen wrote:
houghi wrote:
Appart from 10.0, what beta versions of SUSE have you tested? I have seen none. Could be that they were even worse and completely uninstallable untill the last day.
What about all of 10.1 ? Plenty of betas there. I think I started out with beta3, then 5, 6, 8, 9 and rc1.
Ok, I forgot that people here take each word literealy. houghi -- Nutze die Zeit. Sie ist das Kostbarste, was wir haben, denn es ist unwiederbringliche Lebenszeit. Leben ist aber mehr als Werk und Arbeit, und das Sein wichtiger als das Tun - Johannes Müller-Elmau
houghi wrote:
On Fri, Apr 21, 2006 at 09:10:28AM +0200, Per Jessen wrote:
houghi wrote:
Appart from 10.0, what beta versions of SUSE have you tested? I have seen none. Could be that they were even worse and completely uninstallable untill the last day.
What about all of 10.1 ? Plenty of betas there. I think I started out with beta3, then 5, 6, 8, 9 and rc1.
Ok, I forgot that people here take each word literealy.
Sorry, I didn't mean to offend anyone - what were you trying to say above? /Per Jessen, Zürich
On Fri, Apr 21, 2006 at 10:38:49PM +0200, Per Jessen wrote:
houghi wrote:
On Fri, Apr 21, 2006 at 09:10:28AM +0200, Per Jessen wrote:
houghi wrote:
Appart from 10.0, what beta versions of SUSE have you tested? I have seen none. Could be that they were even worse and completely uninstallable untill the last day.
What about all of 10.1 ? Plenty of betas there. I think I started out with beta3, then 5, 6, 8, 9 and rc1.
Ok, I forgot that people here take each word literealy.
Sorry, I didn't mean to offend anyone - what were you trying to say above?
That previous version did not give us any insight on to what version changes were done at the last moment or how easy or hard Alpha, Beta or RC was to install. Look at it the other way around. If SUSE would not have had released these 10.1XYZ versions, we would just have gotten a (most likely) great distro without knowing about any of the changes and problems that got them there. So I was talking about the whole process of a version before a release. Not only Beta, but also Alpha and Release Candidate. houghi -- Nutze die Zeit. Sie ist das Kostbarste, was wir haben, denn es ist unwiederbringliche Lebenszeit. Leben ist aber mehr als Werk und Arbeit, und das Sein wichtiger als das Tun - Johannes Müller-Elmau
houghi wrote:
Look at it the other way around. If SUSE would not have had released these 10.1XYZ versions, we would just have gotten a (most likely) great distro without knowing about any of the changes and problems that got them there.
I participate in the beta process primarily because it is technically entertaining, and because I run some slightly unusual hardware. I also just like to "do my bit". But in the end I could do quite well without knowing "about any of the changes and problems that got them there". /Per Jessen, Zürich
On Sat, Apr 22, 2006 at 12:15:51AM +0200, Per Jessen wrote:
I participate in the beta process primarily because it is technically entertaining, and because I run some slightly unusual hardware. I also just like to "do my bit". But in the end I could do quite well without knowing "about any of the changes and problems that got them there".
One will not go without the other. Either you wait for a version to come out and perhaps even wait a bit, or you join in on the fun and see the ugliness in all its beauty. Seeing an ugly duckling is one thing. Seeing a beautifull swan is another thing. Seeing it grow from one to another however is the true wonder and beauty. To have been part of it is, for me, more worth that just seeing the beautifull swan. In the end it is not the ugly duckling I remmber, It is the way how it became that beautifull swan. houghi -- Nutze die Zeit. Sie ist das Kostbarste, was wir haben, denn es ist unwiederbringliche Lebenszeit. Leben ist aber mehr als Werk und Arbeit, und das Sein wichtiger als das Tun - Johannes Müller-Elmau
houghi wrote:
One will not go without the other. Either you wait for a version to come out and perhaps even wait a bit, or you join in on the fun and see the ugliness in all its beauty.
Of course - all I said in my OP was that the process has room for improvement, as the amount of work and short cycles are not conducive to quality feedback and/or more testing. /Per Jessen, Zürich
On Friday 21 April 2006 04:06, Boyd Lynn Gerber wrote:
I have most of my machines with only 128 M or 256 M memory.
Something I just wanted to report. Yast became very memory hungry. I didn't notice in the first betas I tried as I had 1GB memory, but due to a module failure now I have only 512MB. But now I see: - starting & closing yast2 sw_single takes 1m 8s. This is an AMD64 3200+ with 512MB, PATA disk (37 MB/sec). I would say decent configuration. - it takes 30% of the memory (I know top is not reliable, but...): 10375 root 18 0 281m 153m 24m S 0.0 30.7 0:10.98 y2base This is a lot. - after installing packages there is a long delay before suseconfig starts to run I think I will not upgrade some SUSE 10 machines. 10 was much better than 9.3 on a Celeron. Time for a new bug report, I would say. Andras -- Quanta Plus developer - http://quanta.kdewebdev.org K Desktop Environment - http://www.kde.org
Andras Mantia schrieb:
Something I just wanted to report. Yast became very memory hungry. I didn't notice in the first betas I tried as I had 1GB memory, but due to a module failure now I have only 512MB. But now I see: - starting & closing yast2 sw_single takes 1m 8s. This is an AMD64 3200+ with 512MB, PATA disk (37 MB/sec). I would say decent configuration. - it takes 30% of the memory (I know top is not reliable, but...): 10375 root 18 0 281m 153m 24m S 0.0 30.7 0:10.98 y2base This is a lot. - after installing packages there is a long delay before suseconfig starts to run
I think I will not upgrade some SUSE 10 machines. 10 was much better than 9.3 on a Celeron.
Time for a new bug report, I would say.
Andras
I have the same problems even though I have 3 gigs of RAM. RC1 is - apart from a whole lot of other nasty things - incredibly slow on my machine, especially yast2.
Danilo Jonas wrote:
I have the same problems even though I have 3 gigs of RAM. RC1 is - apart from a whole lot of other nasty things - incredibly slow on my machine, especially yast2.
Not to contradict anyone, but I've just tried to install 10.1 RC1 on a PII 450MHz with 512M - YaST2 definitely seemed significantly faster in RC1 than in the previous betas. /Per Jessen, Zürich
On 20 Apr 2006 at 14:10, Boyd Lynn Gerber wrote:
Perhaps I'm just whining or venting some steam, but the aggressive release-schedule isn't doing me much good.
I too am very concerned about this and agree 1000%. I wonder the same thing. I do not want to complain but I have real concerns
Basically I want more quality (releases), not more releases. "WONTFIX" or "LATER" doesn't actually solve any problems. Regards, Ulrich
On Thu, 2006-04-20 at 21:01 +0200, Per Jessen wrote:
Perhaps I'm just whining or venting some steam, but the aggressive release-schedule isn't doing me much good.
I don't think a two week schedule makes any sense either - I suspect it is primarily a result of excess energy generated by the advent of openSUSE. Give it a year and the schedule will cool off to something more sensible like once every 3 or 6 months. Also with all the rug/yast/repository work going on with 10.1, there has been essentially no way to get updates for beta releases other than installing the next beta. Hopefully if the update picture stabilizes we can have update servers for beta releases automatically setup when you install and you won't need to know the arcane secret handshake to get an update, and there will be that much less need for ultra frequent releases.
On Thu, Apr 20, 2006 at 06:13:48PM -0400, Tom Horsley wrote:
I don't think a two week schedule makes any sense either - I suspect it is primarily a result of excess energy generated by the advent of openSUSE. Give it a year and the schedule will cool off to something more sensible like once every 3 or 6 months.
Let us look at the lower estimate: 3 months. Say 10.2 development begins may 1st. That would mean 3 months waiting for Alpha 1. Date: August 1st. The next version is November 1st. As february 1st is the release of 10.2 (8 months) The version before that is RC1. So we go from Alpha to RC1 to GM if you take the 3 months period. With a 6 months period, there is 1 release between GMs. While we are at it, can somebody please edit http://en.opensuse.org/Roadmap and the links on the front page? RC3 date and delete the other ones. (W00t! 10.2 Alpha 1 is almost out. No downloads in May?) houghi -- Nutze die Zeit. Sie ist das Kostbarste, was wir haben, denn es ist unwiederbringliche Lebenszeit. Leben ist aber mehr als Werk und Arbeit, und das Sein wichtiger als das Tun - Johannes Müller-Elmau
Tom Horsley wrote:
Also with all the rug/yast/repository work going on with 10.1, there has been essentially no way to get updates for beta releases other than installing the next beta. Hopefully if the update picture stabilizes we can have update servers for beta releases automatically setup when you install and you won't need to know the arcane secret handshake to get an update, and there will be that much less need for ultra frequent releases.
It's a good point, although I'm not sure if updating a beta release is the right way to go. I take the beta-releases to be "take it or leave it", and if I had to rely 100% on the systems I test on, I would not be using the betas at all. I think so far I've managed to get running systems will all the betas and rc1, which is enough for me. I would most likely not be doing an update unless it was to verify a fix. /Per Jessen, Zürich
I can't help noticing a bit of a problem. With a very aggressive release-schedule such as "every 2 weeks", chances of genuine bugs being
1) reported, 2) diagnosed and 3) fixed
in that timeframe are very minimal.
When I say "genuine" I discount typos, apparmor bureacrazy, and other minors that can be dealt with by level 1 support. I've got a couple of open reports (159731, 136742, 159727, 153585), and whilst an aggressive release-schedule is probably a Good Thing(R), it puts a lot of unnecessary strain on testers. For bugs/problems that aren't positively fixed before the next release, it's usually "please install <next release> and see if problem has gone away".
Perhaps I'm just whining or venting some steam, but the aggressive release-schedule isn't doing me much good. There is no aggressive release-schedule. It has always been this way. Only
Am Donnerstag, 20. April 2006 21:01 schrieb Per Jessen: that external testers can see this now, too. But I totally agree with you, that beta testing is a lot of work and having open bugs, sometimes even without comments is not a good thing. Many of the bugfixes you can only test with a fresh install, as you might have done something on your beta, which changes things. So the easiest way is to release betas often. I mean, it's better to install betaN+1 with fixes included, than betaN with additional fixes for testing. I never trust fixes :-) only if the bug has gone with the next release I assume, that it's really fixed. Even then it might come back with the next release. Sometimes a patch does not go in, the developer forgets about it etc. This takes us to another issue: There is also no bugfree distribution (and there will perhaps never be one). So the release depends on the amount of open bugs and the severity of the bugs. -- Mit freundlichen Grüßen, Marcel Hilzinger Linux New Media AG Süskindstr. 4 D-81929 München Tel: +49 (89) 99 34 11 0 Fax: +49 (89) 99 34 11 99
There is also no bugfree distribution (and there will perhaps never be one). So the release depends on the amount of open bugs and the severity of the bugs.
And do not forget: we can see all the bugs now in a list. I can see a bug about fonts in yast that do not work correctly. Fonts I would never use/require, so I would not know the bug. Probably, if we used RC2 for release nobody would directly call it buggy. But when they see the list they all think: Sheiza, this is garbage. It is NOT. It looks slick, it sucks with Zenupdater (but hey, that's gnome-style ;-) ) but for Avarage Joe 10.1 makes live more easy then before. +10 for the dev-team! (and the wallpaper.........so nice :P ) Azerion
On Mon, 2001-01-01 at 19:48 +0100, Azerion wrote: You may want to correct your clock, you're over 5 years behind the rest of us.
There is also no bugfree distribution (and there will perhaps never be one). So the release depends on the amount of open bugs and the severity of the bugs.
And do not forget: we can see all the bugs now in a list. I can see a bug about fonts in yast that do not work correctly. Fonts I would never use/require, so I would not know the bug.
Probably, if we used RC2 for release nobody would directly call it buggy. But when they see the list they all think: Sheiza, this is garbage.
It is NOT. It looks slick, it sucks with Zenupdater (but hey, that's gnome-style ;-) ) but for Avarage Joe 10.1 makes live more easy then before. +10 for the dev-team! (and the wallpaper.........so nice :P )
-- Ken Schneider UNIX since 1989, linux since 1994, SuSE since 1998
On Sun, Apr 23, 2006 at 09:13:49AM -0400, Kenneth Schneider wrote:
On Mon, 2001-01-01 at 19:48 +0100, Azerion wrote:
You may want to correct your clock, you're over 5 years behind the rest of us.
Use positive thoughts. We are 5 years ahead of him. houghi -- Nutze die Zeit. Sie ist das Kostbarste, was wir haben, denn es ist unwiederbringliche Lebenszeit. Leben ist aber mehr als Werk und Arbeit, und das Sein wichtiger als das Tun - Johannes Müller-Elmau
On 1 Jan 2001 at 19:48, Azerion wrote: I guess you've been told meanwhile: Please fix your Date! Ulrich
Op maandag 24 april 2006 08:14, schreef Ulrich Windl:
On 1 Jan 2001 at 19:48, Azerion wrote:
I guess you've been told meanwhile: Please fix your Date!
Ulrich
Allready did ;-) Otherwise all those keycheck go crazy about keys from the future :P Azerion
Marcel Hilzinger wrote:
Many of the bugfixes you can only test with a fresh install, as you might have done something on your beta, which changes things.
I don't see that as so much of a problem - bang in the CD or USB-stick, get the install going, then try out the fix.
So the easiest way is to release betas often. I mean, it's better to install betaN+1 with fixes included, than betaN with additional fixes for testing.
I agree, but the key part being "with fixes included". When no fix has been forthcoming, and another beta has been released, I'm more or less forced to install N+1 even if it does me no good. I do agree that it makes no sense to be working on betaN when N+1 is out, but without fixes, the schedule itself is creating unnecessary work - for testers.
I never trust fixes :-) only if the bug has gone with the next release I assume, that it's really fixed. Even then it might come back with the next release. Sometimes a patch does not go in, the developer forgets about it etc.
I trust the fix if I apply it and the problem is gone, but I don't assume the patch has been applied in the next release, so I do test it again. /Per Jessen, Zürich
participants (12)
-
Andras Mantia
-
Azerion
-
Boyd Lynn Gerber
-
Danilo Jonas
-
houghi
-
Kenneth Schneider
-
Marcel Hilzinger
-
Martin Schlander
-
mchouinard
-
Per Jessen
-
Tom Horsley
-
Ulrich Windl