Hi
I think it would be useful to understand how much contributions of any form in Fedora are sponsored vs volunteers doing it. As a first step can we add in a field in the package database to note down that information?
Rahul
Mike McGrath wrote:
Rahul Sundaram wrote:
Hi
I think it would be useful to understand how much contributions of any form in Fedora are sponsored vs volunteers doing it. As a first step can we add in a field in the package database to note down that information?
Define both.
In a simplied form: Whether you are paid to do some work in Fedora or not.
Rahul
Rahul Sundaram (sundaram@fedoraproject.org) said:
In a simplied form: Whether you are paid to do some work in Fedora or not.
Even if someone is paid to work on some part of Fedora, that doesn't mean they're paid for all their contributions to Fedora. For example, I don't *think* that Red Hat is explicitly paying Jens to maintain Haskell bindings for Gtk+.
Bill
Bill Nottingham wrote:
Rahul Sundaram (sundaram@fedoraproject.org) said:
In a simplied form: Whether you are paid to do some work in Fedora or not.
Even if someone is paid to work on some part of Fedora, that doesn't mean they're paid for all their contributions to Fedora. For example, I don't *think* that Red Hat is explicitly paying Jens to maintain Haskell bindings for Gtk+.
This is precisely why having a field with this information per package in the package database is useful.
Rahul
Rahul Sundaram wrote:
Bill Nottingham wrote:
Rahul Sundaram (sundaram@fedoraproject.org) said:
In a simplied form: Whether you are paid to do some work in Fedora or not.
Even if someone is paid to work on some part of Fedora, that doesn't mean they're paid for all their contributions to Fedora. For example, I don't *think* that Red Hat is explicitly paying Jens to maintain Haskell bindings for Gtk+.
This is precisely why having a field with this information per package in the package database is useful.
This sounds like bloat to me. While this is something we could do, is it really something we should?
-Mike
Mike McGrath wrote:
Rahul Sundaram wrote:
Bill Nottingham wrote:
Rahul Sundaram (sundaram@fedoraproject.org) said:
In a simplied form: Whether you are paid to do some work in Fedora or not.
Even if someone is paid to work on some part of Fedora, that doesn't mean they're paid for all their contributions to Fedora. For example, I don't *think* that Red Hat is explicitly paying Jens to maintain Haskell bindings for Gtk+.
This is precisely why having a field with this information per package in the package database is useful.
This sounds like bloat to me. While this is something we could do, is it really something we should?
I think it is useful to understand the nature of contributions. It has some up several times before in various discussions that we had in the past about strength of the Fedora community and vendor participation.
Rahul
Rahul Sundaram wrote:
Mike McGrath wrote:
Rahul Sundaram wrote:
Bill Nottingham wrote:
Rahul Sundaram (sundaram@fedoraproject.org) said:
In a simplied form: Whether you are paid to do some work in Fedora or not.
Even if someone is paid to work on some part of Fedora, that doesn't mean they're paid for all their contributions to Fedora. For example, I don't *think* that Red Hat is explicitly paying Jens to maintain Haskell bindings for Gtk+.
This is precisely why having a field with this information per package in the package database is useful.
This sounds like bloat to me. While this is something we could do, is it really something we should?
I think it is useful to understand the nature of contributions. It has some up several times before in various discussions that we had in the past about strength of the Fedora community and vendor participation.
So I should log in to the package database and click yes I'm being paid to do this or no I'm know being paid to do this? Whats the default? Where can I find these various discussions?
-Mike
Mike McGrath wrote:
So I should log in to the package database and click yes I'm being paid to do this or no I'm know being paid to do this? Whats the default?
Let's say you can have two fields
Sponsored: Yes/No. Default: No (Since most of the packages in the repository are maintained voluntarily and we expect the percentage to increase)
Sponsored by: <Empty>. Inactive if no.
Where can I find these various discussions?
No public archives. Most of these were on phone or in person. Fundamentally it is another metric that determines the health of the community (users and contributors) and our understanding of them, like the number of users (Statistics page) or what applications they run usually (mugshot app stats)
Rahul
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Rahul Sundaram wrote:
Mike McGrath wrote:
So I should log in to the package database and click yes I'm being paid to do this or no I'm know being paid to do this? Whats the default?
Let's say you can have two fields
Sponsored: Yes/No. Default: No (Since most of the packages in the repository are maintained voluntarily and we expect the percentage to increase)
Sponsored by: <Empty>. Inactive if no.
Where can I find these various discussions?
No public archives. Most of these were on phone or in person. Fundamentally it is another metric that determines the health of the community (users and contributors) and our understanding of them, like the number of users (Statistics page) or what applications they run usually (mugshot app stats)
Rahul
Then would you also want to know by whom a person may be paid to make the contribution? Does this go per change? Would you suppose I click the sponsored button if I'm being paid for package X most of the time but John Doe has volunteered to provide me with a patch I update the package with, is that sponsored or not sponsored? Isn't this more appropriately addressed with a questionair you request your contributors to fill out (wiki poll)?.
Just my thoughts.
Kind regards,
Jeroen van Meeuwen - -kanarip
Jeroen van Meeuwen wrote:
Then would you also want to know by whom a person may be paid to make the contribution?
I dont think we really need that level of information
Does this go per change? Would you suppose I click the
sponsored button if I'm being paid for package X most of the time but John Doe has volunteered to provide me with a patch I update the package with, is that sponsored or not sponsored?
Per patch would be more accurate but too intrusive. I prefer to have metrics that give you information with some amount of fuzziness. We know that such stats are not entirely accurate even it it is done on a per patch basis but it does give you a basic understanding. That is all I am aiming for.
Isn't this more appropriately
addressed with a questionair you request your contributors to fill out (wiki poll)?.
I had already thought about that. Polls are asynchronous and has to be done on a periodical basis. Having this information as part of infrastructure is going to be successful in giving you more complete metrics.
Rahul
Rahul Sundaram wrote:
Mike McGrath wrote:
So I should log in to the package database and click yes I'm being paid to do this or no I'm know being paid to do this? Whats the default?
Let's say you can have two fields
Sponsored: Yes/No. Default: No (Since most of the packages in the repository are maintained voluntarily and we expect the percentage to increase)
What if I maintain packages for work, and some for personal use?
How do you count work done on packages owned by other people?
This seems like an inaccurate and misleading metric. Meanwhile, we have many *far* more relevant things that we could measure that we haven't done so yet like: https://www.redhat.com/archives/fedora-infrastructure-list/2007-March/msg002...
Warren Togami wtogami@redhat.com
Warren Togami wrote:
Rahul Sundaram wrote:
Mike McGrath wrote:
So I should log in to the package database and click yes I'm being paid to do this or no I'm know being paid to do this? Whats the default?
Let's say you can have two fields
Sponsored: Yes/No. Default: No (Since most of the packages in the repository are maintained voluntarily and we expect the percentage to increase)
What if I maintain packages for work, and some for personal use?
Are you even reading the thread? I have already answered that.
How do you count work done on packages owned by other people?
Also answered in Toshio's mail.
This seems like an inaccurate and misleading metric.
This seems more like you jumped into a thread without reading it completely.
Rahul
On Tue, 2007-04-24 at 16:20 -0500, Mike McGrath wrote:
Rahul Sundaram wrote:
Bill Nottingham wrote:
Rahul Sundaram (sundaram@fedoraproject.org) said:
In a simplied form: Whether you are paid to do some work in Fedora or not.
Even if someone is paid to work on some part of Fedora, that doesn't mean they're paid for all their contributions to Fedora. For example, I don't *think* that Red Hat is explicitly paying Jens to maintain Haskell bindings for Gtk+.
This is precisely why having a field with this information per package in the package database is useful.
This sounds like bloat to me. While this is something we could do, is it really something we should?
I'm usually of the opinion that more information is better as long as it doesn't compromise someone's privacy. If this is optional information it probably wouldn't violate that principle. I could see this kind of data being useful for things like:
1) Estimating how much of an investment companies besides Red Hat have in Fedora. 2) Having figures on community contributors to post as part of a reply to "Fedora is RHEL Beta".
Here's a stab at how this would work:
A package could be maintained by multiple people. One might be paid to do work on it but another would not. So this needs to be part of a person/package linking table. We have a table that links person->package-in-collection so that might be appropriate.
We probably want to track something more complex than a boolean; sponsored by (self|Red Hat|Dell|Pogo|School District).
So adding a nullable field to the personpackagelisting table that references a list of companies (updatable by users) is probably a good first approximation. (Another interesting thing would be to categorize the organization so we could see which sectors are paying people to work on Fedora but the more collected information, the more work a contributor would have to do to enter it.)
One problem with this is that the owner is currently embedded within the packagelisting (this made sense as every package needs to have one packager that is the owner.) The best way to change this is probably to change this to create a personpackagelisting entry for every owner and have the owner field in the packagelisting table point at that personpackagelisting entry.
Rahul, if you'd like to put this on the list at the bottom of:: http://fedoraproject.org/wiki/Infrastructure/PackageDatabase
I can look at it in more detail later. It won't hit before F7 unless someone else wants to do the work, though.
-Toshio
Toshio Kuratomi wrote:
Rahul, if you'd like to put this on the list at the bottom of:: http://fedoraproject.org/wiki/Infrastructure/PackageDatabase
I can look at it in more detail later. It won't hit before F7 unless someone else wants to do the work, though.
Added.
Rahul
Toshio Kuratomi (a.badger@gmail.com) said:
I'm usually of the opinion that more information is better as long as it doesn't compromise someone's privacy. If this is optional information it probably wouldn't violate that principle. I could see this kind of data being useful for things like:
- Estimating how much of an investment companies besides Red Hat have
in Fedora.
However, if it's only optional...
- Having figures on community contributors to post as part of a reply
to "Fedora is RHEL Beta".
It leaves out so many ways that people can be part of the community...
... write docs for Fedora ... do artwork for Fedora ... do testing of Fedora ... port Fedora to other platforms ... contribute patches to Fedora ... do upstream development with/on Fedora ... maintain systems for Fedora
I'm not saying that collecting these metrics, or things like Warren's different-levels-of-packaging proposal aren't something that can be useful. But I'd like us to think how we can somehow collect stats that somehow encompass the entirety of someone's contributions.
Bill, off in pie-in-the-sky land
On Tue, 2007-04-24 at 22:53 -0400, Bill Nottingham wrote:
Toshio Kuratomi (a.badger@gmail.com) said:
I'm usually of the opinion that more information is better as long as it doesn't compromise someone's privacy. If this is optional information it probably wouldn't violate that principle. I could see this kind of data being useful for things like:
- Estimating how much of an investment companies besides Red Hat have
in Fedora.
However, if it's only optional...
- Having figures on community contributors to post as part of a reply
to "Fedora is RHEL Beta".
It leaves out so many ways that people can be part of the community...
... write docs for Fedora ... do artwork for Fedora ... do testing of Fedora ... port Fedora to other platforms ... contribute patches to Fedora ... do upstream development with/on Fedora ... maintain systems for Fedora
I'm not saying that collecting these metrics, or things like Warren's different-levels-of-packaging proposal aren't something that can be useful. But I'd like us to think how we can somehow collect stats that somehow encompass the entirety of someone's contributions.
Oh yeah! Now you're talking :-). I would really love to be able to hit a web page enter someone's name and get a list of all their contributions to Fedora. Some nice summaries and graphs (% patches accepted into the package; % Fedora contribution in each subproject,...) It would be an interesting cross with social networking to see how projects you work on intersect with projects that other Fedora Contributors work on. Hmmm... I work on the Turbogears package and Luke Macken works on TurboGears. Luke works on Fedora Infrastructure projects. Wow, Infrastructure is using TurboGears all over the place! I think I should sign up to help out there!
-Toshio
On Tue, 2007-04-24 at 20:35 -0700, Toshio Kuratomi wrote:
On Tue, 2007-04-24 at 22:53 -0400, Bill Nottingham wrote:
It leaves out so many ways that people can be part of the community...
[snip list]
Oh yeah! Now you're talking :-).
I agree, *very* interesting. Obviously, we can pursue Rahul's idea right now (well, after F7), while we can talk about how to do the pie-in-the-sky idea.
One thing we can hope for is that, for most contributors, there is a consistent default. Then even if we ask for every e.g. Wiki page save, "Was this for-pay or for-personal?" the unique default for each person is going to most often be the right one.
With FAS, we have the backbone needed -- all services talk with the same account system, so all services can derive default and follow-on values, as well as write back new values when they come up. When contributing using a service such as CVS or Plone, your default is picked up, your optional values are a tab-completion away, and there is always a way to force a new value.
But what these all call for is a social convention adjustment. Just as it takes effort to get a group of developers who don't use changelogs well to remember to include '-m "Something meaningful, damnit!"' for every commit. We would be asking, for example, add a keyword if it is *not* your default (cvs ci -m "Something meaningful; *sponsor=Big Company*").
So the first question is really, how do we get buy-in from across the project that the extra hassle is worth the value it brings Fedora to know where we stand at all times?
- Karsten
Karsten Wade (kwade@redhat.com) said:
So the first question is really, how do we get buy-in from across the project that the extra hassle is worth the value it brings Fedora to know where we stand at all times?
What's the point of having the data? What are we intending to do with it? What does it buy us?
You need to know that before you know if it's worth the hassle. It's essentially navel-gazing sort of numbers, that may not have a *direct* effect on any particular aspect of the project in the short term.
Bill
Bill Nottingham wrote:
Karsten Wade (kwade@redhat.com) said:
So the first question is really, how do we get buy-in from across the project that the extra hassle is worth the value it brings Fedora to know where we stand at all times?
What's the point of having the data? What are we intending to do with it? What does it buy us?
You need to know that before you know if it's worth the hassle. It's essentially navel-gazing sort of numbers, that may not have a *direct* effect on any particular aspect of the project in the short term.
My goal was marketing. Showing community and multi vendor participation in Fedora has a very direct effect on perceptions on how things evolved in Fedora.
There are other benefits like Toshio said but this alone makes it worth the effort IMO.
Rahul
infrastructure@lists.fedoraproject.org