It appears that octave has been removed from Fc4-test.
I find this to be rather unfortunate, as I often cite octave as one of the fantastic professional grade tools that are just 'there' on Fedora. I don't know that being able to say that open office contains dozens of languages I don't speak carries the same importance as having octave.
If you use the metric of 'used by most users' and apply it to all the packages, you could probably get the distribution down to a single CD, two for sure.
Historically packages in extras have not been as well maintained... and there isn't an easy way to download isos and install 'everything' (where everything includes extras) from physical media, at least not that I'm aware of.
If media count is a primary motivator, why not split the distribution up so that there is a 'base' which fits on one CD, 'complete' which takes no more than one or two DVDs and includes all packages that are expected to be of a high quality, and 'extras' ... which has everything else?
On Fri, Mar 18, 2005 at 03:06:15PM -0500, seth vidal wrote:
Historically packages in extras have not been as well maintained...
I'd love to see some proof of that. ANY proof of that, please.
Not to mention that Extras hasn't existed long enough for there to really be a history...
On Fri, 2005-03-18 at 15:09 -0500, Chuck R. Anderson wrote:
On Fri, Mar 18, 2005 at 03:06:15PM -0500, seth vidal wrote:
Historically packages in extras have not been as well maintained...
I'd love to see some proof of that. ANY proof of that, please.
Not to mention that Extras hasn't existed long enough for there to really be a history...
If you look at the maintenance of packages in Fedora Extras 3 you should see a good history there. Not only of active packagers but also of recent builds of things.
so if you go only by the available history, I think things look pretty damned good.
-sv
On Fri, 18 Mar 2005 15:09:47 -0500, Chuck R. Anderson cra@wpi.edu wrote:
On Fri, Mar 18, 2005 at 03:06:15PM -0500, seth vidal wrote:
Historically packages in extras have not been as well maintained...
I'd love to see some proof of that. ANY proof of that, please.
Not to mention that Extras hasn't existed long enough for there to really be a history...
Gah. I should have careful with my words... What I intended to convey is that packages included in the core distribution tend to be better maintained. This wasn't intended to be a specific slam against extras, and I'm disappointed that this one point from my post is drawing attention to the exclusion of the others.
The statement that in-core packages get more attention is pretty basic.. As a part of the core installation they get installed with a lot of other cruft along with an 'everything' installed and as a result they get a fair amount of casual 'whats this' use beyond the use that packages people must seek out and install by name get. As a part of the core distribution build problems in such package can become roadblocks to scheduled releases where being in extras is more of a 'if it's there it's there', it changes the priority of a package.
I believe that one of the huge advantages of the free software licensing model for the user is that there are greatly reduced barriers to using the full professional grade tool rather than some toy... By excluding such tools we increase the cost to the user from just being the learning curve to locating and installing the tool, this is a substantial change for those of us with plenty of disk space but who are often mobile and away from a fast network connection.
Of course, there are reasons not to include every piece of software out there... For example, it's nice to have a base distro that fits on a single piece of media. It's also good to only include packages with a distro if there is going to be some commitment to the quality of the packages. These are fine reasons, but if we are going to use them lets carry them to their logical extents, ... make a base distro that fills a single piece of media, and an extension that carries everything meeting the quality requirements with little regard to the additional size. An other set of breaks ends up annoying people with different usage patterns than you for no good reason (why do I have to goto extras to get octave, which I have much use for... while OpenOffice is included when it's SO much larger and I have absolutely no use for it?... you can't really say popularity because package usage is likely a long tail distribution and there are lots of packages that individually have little usage but collectively comprise a majority of the used applications)
On Fri, Mar 18, 2005 at 04:00:22PM -0500, Gregory Maxwell wrote:
Gah. I should have careful with my words... What I intended to convey is that packages included in the core distribution tend to be better maintained. This wasn't intended to be a specific slam against
While this should be taken with the normal caution around forward-looking statements, I don't think that this will necessarily be the case, especially for special-domain programs like Octave. With FC-mumble-mumble-5, installing packages from Extras will be as easy as installing packages from Core. And, since they can be maintained by someone specialized in the particular field and well-versed in the actual program -- maybe someone who *just* maintains that package with no other Fedora workload -- I expect many Extras packages to get *better* maintenance than they would in Core.
On Fri, 18 Mar 2005 17:42:54 -0500, Matthew Miller mattdm@mattdm.org wrote:
While this should be taken with the normal caution around forward-looking statements, I don't think that this will necessarily be the case, especially for special-domain programs like Octave. With FC-mumble-mumble-5, installing packages from Extras will be as easy as installing packages from Core. And, since they can be maintained by someone specialized in the particular field and well-versed in the actual program -- maybe someone who *just* maintains that package with no other Fedora workload -- I expect many Extras packages to get *better* maintenance than they would in Core.
Achieving that without ending up with FC-mumble coming out without an x (insert package in extras) that will work on it being available for weeks will be quite an accomplishment. Without this level of assurance we're asking users that depend on 'extras' packages to undertake a difficult additional step of comparing the packages they use against the available extras prior to upgrading (since compatibility is broken from time to time).
As far as installing ease, unless you mean that extras will be available as an additional DVD image, and the installer prompts the user for additional disks so they can be added to the package selection (and the bloat-me-harder everything install) then we will have not achieved the same ease as packages in the base install.
If extras do indeed turn out to be so well maintained and easy to install that my concerns turn out warrentless, then we really should be moving the bulk of the distribution into extras.
On Fri, Mar 18, 2005 at 09:11:16PM -0500, Gregory Maxwell wrote:
On Fri, 18 Mar 2005 17:42:54 -0500, Matthew Miller mattdm@mattdm.org wrote: Achieving that without ending up with FC-mumble coming out without an x (insert package in extras) that will work on it being available for weeks will be quite an accomplishment. Without this level of assurance
Um, I think I need either some more punctuation in the above sentence or some more coffee -- Huh? :)
we're asking users that depend on 'extras' packages to undertake a difficult additional step of comparing the packages they use against the available extras prior to upgrading (since compatibility is broken from time to time).
Packages have always been dropped from the distro. Moving them to extras instead of dropping them seems like a huge win here.
As far as installing ease, unless you mean that extras will be available as an additional DVD image, and the installer prompts the user for additional disks so they can be added to the package selection (and the bloat-me-harder everything install) then we will have not achieved the same ease as packages in the base install.
Yes, I think this is a goal.
If extras do indeed turn out to be so well maintained and easy to install that my concerns turn out warrentless, then we really should be moving the bulk of the distribution into extras.
And yes, I'd like to see this too. "Core" should be just that.
On Fri, 18 Mar 2005 16:00:22 -0500, Gregory Maxwell wrote:
On Fri, 18 Mar 2005 15:09:47 -0500, Chuck R. Anderson cra@wpi.edu wrote:
On Fri, Mar 18, 2005 at 03:06:15PM -0500, seth vidal wrote:
Historically packages in extras have not been as well maintained...
I'd love to see some proof of that. ANY proof of that, please.
Not to mention that Extras hasn't existed long enough for there to really be a history...
Gah. I should have careful with my words... What I intended to convey is that packages included in the core distribution tend to be better maintained. This wasn't intended to be a specific slam against extras, and I'm disappointed that this one point from my post is drawing attention to the exclusion of the others.
Here you did it again. Generalisation, such as "packages included in the core distribution tend to be better maintained", is bad. First you should define how you measure quality of maintenance.
The statement that in-core packages get more attention is pretty basic.. As a part of the core installation they get installed with a lot of other cruft along with an 'everything' installed and as a result they get a fair amount of casual 'whats this' use beyond the use that packages people must seek out and install by name get.
How many users perform "everything" installations and try only a small fraction of the installed packages? How many users don't even install Fedora Core Updates? How many report bugs if they don't have special interest in the software? Compare that with users who are in search of a small set of specific extra applications, e.g. in the freshmeat.net directory, find them as recent rpms in Fedora Extras, and would not give them a try if they had to build them from source code. With the availability of ready-to-use packages in Fedora Extras and with their interest in the software, they become dedicated users ("power-users"). And it's these users who also report bugs or seek for communication with upstream developers.
As a part of the core distribution build problems in such package can become roadblocks to scheduled releases where being in extras is more of a 'if it's there it's there', it changes the priority of a package.
In case of build problems or bugs, what makes you so sure that packages in Fedora Extras don't gain special attention from their maintainers? Maybe the maintainer of a package in FE is the author of the software? Maybe the maintainer of a package in FE can focus on a single package, because he/she only maintains a single package in FE?
On Fri, 2005-03-18 at 16:00 -0500, Gregory Maxwell wrote:
Historically packages in extras have not been as well maintained...
I'd love to see some proof of that. ANY proof of that, please.
Not to mention that Extras hasn't existed long enough for there to really be a history...
Gah. I should have careful with my words... What I intended to convey is that packages included in the core distribution tend to be better maintained. This wasn't intended to be a specific slam against extras, and I'm disappointed that this one point from my post is drawing attention to the exclusion of the others.
I see the faith you have in Red Hat Engineering, and with Core, but not with Extras for some reason? Well, if you look at the cvs-commits list, you'll notice quite a lot of Red Hat employees pushing packages into Extras
So this idea of "better maintained" is by far a myth
The statement that in-core packages get more attention is pretty basic.. As a part of the core installation they get installed with a lot of other cruft along with an 'everything' installed and as a result they get a fair amount of casual 'whats this' use beyond the use that packages people must seek out and install by name get. As a part of the core distribution build problems in such package can become roadblocks to scheduled releases where being in extras is more of a 'if it's there it's there', it changes the priority of a package.
And the Core distribution is trying to push the "one good utility to do the job". Granted that octave has no replacement in Core, but it also isn't part of the Core uses of using a Linux desktop or server, is it?
Maybe at some stage, when we get FC5 or so, we can have Extras organised in different forms, much like system-config-packages, or groups as you may. "Fedora Extras Scientific", and pull that group in and have fun
Oh wait, we already have that with groups in yum :)
(but on a more serious note, we should be getting extras sorted that way too at some stage)
I believe that one of the huge advantages of the free software licensing model for the user is that there are greatly reduced barriers to using the full professional grade tool rather than some toy... By excluding such tools we increase the cost to the user from just being the learning curve to locating and installing the tool, this is a substantial change for those of us with plenty of disk space but who are often mobile and away from a fast network connection.
Hopefully Extras ends up on a DVD or a few CDs as well. Its on the todo list, afaik
Why not maintain octave in Extras, if no one else has already taken it? And make sure its the same high quality stuff you talk about :)
Maybe at some stage, when we get FC5 or so, we can have Extras organised in different forms, much like system-config-packages, or groups as you may. "Fedora Extras Scientific", and pull that group in and have fun
Oh wait, we already have that with groups in yum :)
(but on a more serious note, we should be getting extras sorted that way too at some stage)
absolutely, in fact, this is something I think anyone and everyone should start submitting groups of.
submit a comps.xml-compliant group to bugzilla logged against: product: fedora-extras component: general
We need to get extras organized better and I think that this way is a good way to start.
Anyone object?
-sv
-
Chuck Anderson -
Colin Charles -
Gregory Maxwell -
Matthew Miller -
Michael Schwendt -
seth vidal