I am a consumer of Fedora Updates–updated software shipped for supported Fedora releases. I greatly appreciate the work so many package (software) maintainers do in Fedora to make these updates available. I confess to rarely, if ever, being able to tell if I need the update or if having installed the update, my system runs any different a month or so after the GA release.
If this is true, would it be better if our amazing maintainers focused less on updates to stable releases and more on the release under development? Would we gain more by focusing our time and resources on updates on a certain group of packages? Would our releases be more stable if we deferred more enhancements and minor bug fixes to the release under development?
I’m not talking about security updates. I’m all for closing security holes. I’m talking about your garden variety enhancement or bug fix.
Of the hundreds of megabytes of downloaded packages–probably gigabytes over the life of a supported release–how many of those updates did I definitely need because they solved a bug I was actively experiencing? What is it with my fascination and urge to compulsively run ‘yum update’ or click on the ‘updates are available for your system’ icon all the time the way some people check their email?
The only answer I can come up with is that it is tied to an unvalidated hope that “running the latest and greatest” software will result in fewer problems.
Fedora is generally problem free on my workstation. I have no complaints. My notebook is another story.
I have a Dell XPS M1130 that is ten months old. It has Intel graphics and networking. It is currenly running Fedora 11 with the latest updates. Travelling with it for FUDCon Toronto was an unpleasant experience. The following problems have happened to me for most of Fedora 11:
- After suspending and resuming multiple times the machine freezes and requires a hard reset
- Resuming from suspend the wireless network interface completely disappears and only comes back after rebooting
- Resuming from suspend, NetworkManager disappears and only comes back after rebooting (restarting the service doesn’t work)
- Hibernate works most of the time except when it doesn’t
This basic functionality that works flawlessly on my wife’s three year old Dell Inspiron (running Windows XP) has failed more times than I can count running Fedora. I’m guessing these are kernel bugs specific to my particular hardware, but overall I would expect Fedora 11 to be more mature by now, particularly on this relatively established hardware by a major vendor.
I find myself wondering today if I just need “a bigger set of updates”–update to Fedora 12–to solve these problems. Something about the approach of “hopefully some more updates will fix these very basic problems” seems wrong. Shouldn’t I be able to run a “stable” Fedora release on a notebook to accomplish general productivity tasks without having to reboot all the time?
A lot of people in Fedora are probably tired of hearing about “defining a Target Audience” for Fedora, but to me issues like the ones I mention above point directly to it.
Is this an acceptable user experience? Is this a compelling user experience to attract new users? Would I want someone less experienced with Linux to have this be their first impression of Fedora? For those that would argue that my problems are “someone else’s problem and the result of an unfixed bug upstream” I still see this as a problem of the Fedora brand. Basic functionality that does not work on mainstream hardware tarnishes our brand.
What am I missing?
December 25, 2009 at 8:53 am
It would be great if everything worked perfectly on release day. However, in practice, bug-free software doesn’t exist, so bugfix updates are very much needed!
As for enhancement updates, those are the very updates you would actually notice even if you didn’t have any issues with the software, and in a positive way, so I don’t see why we should stop doing them. Of course we need to be careful not to break or regress things, but many enhancement updates are perfectly safe.
So, to answer your questions (or contradict them if they were rhetorical 😉 ):
> If this is true,
It’s not, so…
> would it be better if our amazing maintainers focused less on updates to stable releases and more on the release under development?
… no. 🙂
I think it is very important to actively maintain our current releases. Most people don’t want to run Rawhide with all the breakage that goes with it, nor wait 6 months for any improvement. I know I don’t want either option. (Oh, and I don’t want to have to wait a month or even a week for my updates either as has been floated around. We really need frequent update pushes! People who don’t want to update daily are free to only run their updates weekly or even monthly.)
> Would we gain more by focusing our time and resources on updates on a certain group of packages?
What would that group of packages be? IMHO, it’s “everything”. 🙂
> Would our releases be more stable if we deferred more enhancements and minor bug fixes to the release under development?
I doubt it. Our updates are normally for the better, not for the worse. We may need more effective QA to catch the odd ones out, but in principle the updates are fine.
December 22, 2009 at 2:06 pm
I’m also one of those users who update everything all of the time in some vain hope that one or other of the updates might be useful … 🙂
I’m using a lenovo (/ IBM) ThinkPad T61 with nvidia graphics, and when I was using Fedora 11 this meant that I could chose to have working multiheaded, compositing or suspend/resume, but not all of them, by choosing which driver to use (nv, nouveau or proprieatry nvidia). The proprietory nvidia generally worked best, but some times after a kernel update it didn’t work at all …
Now I’m using Fedora 12 and all of the above works, using the nouveau driver.
December 22, 2009 at 8:50 am
I have similar suspend/resume problems with my common hardware (intel i965gm chipset, doesn’t resume at all). I used Fedora 11, and after I’ve updated to kernel 2.6.30, suspend/resume started to fail. I hoped that the problem will go away when I upgrade to F12. I had no luck, resume is still broken. It happens, but what makes me very disappointed, is the maintainers response to the reported bug mentioning this problem. I mean no response. I’m not a linux expert, just an experienced user, but I would like to help by providing the needed information. But nobody cares. What I see is that nobody is interested on the fedora side. I think that this is a serious damage on the Fedora brand.
December 22, 2009 at 12:49 am
My main problem is when during a release cycle the hardware support changes… sure, all is well when after an update a device start working, but is depressing when it stop working. And it happens.
December 21, 2009 at 10:50 pm
This is a long reply, sorry for that.
I apply updates normally within 24 hours of release, and while the changes I see are rarely significant that isn’t always the case. I ‘hawk’ my main system and my own experience is that every set of updates is a crap shoot on what effect they will have.
Cherry picking the updates has never been the fedora way and I update to much I know. Most updates are in fact not worth the effort to download. I download them in the (unrealistic) hope that the updates are tuning or stabilization enhancements.
I submit the reason F11 runs like it does with no apparent relief to your issues is a direct result Fedora’s development philosophy. Released software is dead software. An even greater focus on the development release would only cause the ‘supported’ release to be even more flaky with no possibility of correction.
“Shouldn’t I be able to run a “stable” Fedora release on a notebook to accomplish general productivity tasks without having to reboot all the time?”
The current answer to that question is no. Fedora does not offer a “stable” release, only a testing release with less testing required by the user as the version ages.
That is in no way meant to be a smartarse response just a factual one based on reality not words on a wiki page.
“A lot of people in Fedora are probably tired of hearing about “defining a Target Audience” for Fedora,…”
I’m not! I will go further and say I want to hear more about it. You are spot on when you say all of this is directly related to the “Target Audience” question. I was under the impression that this question was the be resolved by end of FUDcon.
Right now Fedora is: A laptop centric distribution that serves as a field test bed for mostly but not exclusively new desktop technologies for possible inclusion in to RHEL.
The target Audience is: Experienced Linux users and developers that are interested in becoming contributors and willing to run what is essentially a testing distribution.
Fedora the Project preforms that last part very successfully IMHO. Fedora allows me the opportunity to contribute with little to no restriction. Running fedora allows me to continue to contribute.
Fedora the distribution on the other hand is starting to reach a point of diminished returns. With each release it fits my needs as a user less and less out of the box.
I’ve all but already made up my mind that F13 wont be installed on anything with the exception of my test machine, even after release.
So to answer the questions
“Is this an acceptable user experience?”
For me personally, no it is not.
“Is this a compelling user experience to attract new users?”
I don’t advise users new or experienced to use fedora, as much as I would like to there is no compelling reason to choose fedora over some of the other offerings out there.
The biggest thing fedora has going for it IMO is the ability contribute. It has no technical merits that would make it stand out.
Not the most cutting edge (this is another way of saying “Not the most technically advanced.”).
Not the most stable.
Not the most used.
Not the fastest.
Not the freest (is that even a word?).
Having no defined “target audience” for so long has had advantages. People that probably shouldn’t run fedora have because of that ambiguity. It also let fedora become the ‘sybil’ of Linux Distro’s; or a “Jack of all trades and master of none”.
Fedora can’t be everything to everyone. That being the case, no matter what target audience is finally defined, it will undoubtedly cost the distribution users, and the project members as well. It couldn’t possibly be otherwise.
If this makes the project stronger and the distro better, then the cost is worth the reward.
“What am I missing?”
Nothing.
We are drinking our own koolaid and starting to not like the taste.
As a fedora user and contributor I would really like to know where this boat is headed. I eagerly await the answer.
I promise to never make such a long reply again =)