Here's the list of items currently on the agenda for tomorrow's (20090306) FESCo meeting.
57 Make stronger policy/guideline that ABI/API/soname breakage should always be announced fedora-devel-announce (not f-d-l) 67 How to deal with Requires.privates in pkg-config 70 provenpackager guidelines & reseed 79 Lenient hash check for noarch subpackages 89 Select Fedora representative for LSB 96 Document "Rawhide cannot go backwards" rule 97 rpm 4.7 in Fedora 11 65 Fedora Creative Commons Content repository
For more complete details, please visit each individual ticket. The report of the agenda items can be found at https://fedorahosted.org/fesco/report/9
If you would like to add something to this agenda, you can reply to this e-mail, file a new ticket at https://fedorahosted.org/fesco, e-mail me directly, or bring it up at the end of the meeting, during the open floor.
FPC will have two items for consideration:
https://fedoraproject.org/wiki/PackagingDrafts/ExplicitRequires https://fedoraproject.org/wiki/Troublesome_source_URL_packaging_guideline_dr...
Sorry I've been jammed up with the minutes this week; they should be out soon.
- J<
On Thu, Mar 5, 2009 at 3:47 PM, Jason L Tibbitts III tibbs@math.uh.edu wrote:
Sorry I've been jammed up with the minutes this week; they should be out soon.
Not a problem, I've added these to the agenda.
Jon Stanley (jonstanley@gmail.com) said:
If you would like to add something to this agenda, you can reply to this e-mail, file a new ticket at https://fedorahosted.org/fesco, e-mail me directly, or bring it up at the end of the meeting, during the open floor.
Are we reviewing the feature list for completeness w.r.t. feature freeze?
Bill
On Thu, Mar 05, 2009 at 03:47:45PM -0500, Bill Nottingham wrote:
Jon Stanley (jonstanley@gmail.com) said:
If you would like to add something to this agenda, you can reply to this e-mail, file a new ticket at https://fedorahosted.org/fesco, e-mail me directly, or bring it up at the end of the meeting, during the open floor.
Are we reviewing the feature list for completeness w.r.t. feature freeze?
<aol> me too </aol>
I cancelled some stuff so I can be at the meeting.
Rich.
Bill Nottingham wrote:
Jon Stanley (jonstanley@gmail.com) said:
If you would like to add something to this agenda, you can reply to this e-mail, file a new ticket at https://fedorahosted.org/fesco, e-mail me directly, or bring it up at the end of the meeting, during the open floor.
Are we reviewing the feature list for completeness w.r.t. feature freeze?
Bill
I will file a ticket later tonight or first thing tomorrow morning with a list of features pages where the feature page is stale or if I can't determine if the feature "testable".
I will not be able to attend the meeting, but will read the scroll back later.
John
On 03/05/2009 10:44 PM, Jon Stanley wrote:
Here's the list of items currently on the agenda for tomorrow's (20090306) FESCo meeting.
57 Make stronger policy/guideline that ABI/API/soname breakage should always be announced fedora-devel-announce (not f-d-l) 67 How to deal with Requires.privates in pkg-config 70 provenpackager guidelines & reseed 79 Lenient hash check for noarch subpackages 89 Select Fedora representative for LSB 96 Document "Rawhide cannot go backwards" rule 97 rpm 4.7 in Fedora 11 65 Fedora Creative Commons Content repository
For more complete details, please visit each individual ticket. The report of the agenda items can be found at https://fedorahosted.org/fesco/report/9
If you would like to add something to this agenda, you can reply to this e-mail, file a new ticket at https://fedorahosted.org/fesco, e-mail me directly, or bring it up at the end of the meeting, during the open floor.
It would be nice if you can allocate 3' to refine the non-responsive maintainer + closed ACL policy. See https://www.redhat.com/archives/fedora-devel-list/2009-March/msg00281.html and the thread leading to it. A month ago I wanted myself to fix gdal and its broken deps but I could not due to the closed ACL.
On 05.03.2009 21:44, Jon Stanley wrote:
57 Make stronger policy/guideline that ABI/API/soname breakage should always be announced fedora-devel-announce (not f-d-l)
Why not simply mail f-d-l as well as all the maintainers of packages that are affected (and not fedora-devel-announce!)? A simple command line script is able to do that in case MTA's like sendmail or postfix are properly configured. That script likely could write a comma separated list that people can cut-n-paste to Thunderbird, Kmail and other MUA's if needed.
Reason: I for one slowly start to get annoyed by the slowly increasing traffic on fedora-devel-announce that clutters my inbox. Thus I started to consider to move mails from that list into some IMAP folder automatically, and that is exactly what we didn't want people to do when fedora-devel-announce was created. I think we said something like "less then 10 mails a month" back when we created it, but I could not find that on a quick google search :-/
CU thl
P.S.: I on purpose comment here on the list and not in the ticket where this comment would be hidden by most of those that will be are affected by the change.
On Fri, 2009-03-06 at 07:03 +0100, Thorsten Leemhuis wrote:
Why not simply mail f-d-l as well as all the maintainers of packages that are affected (and not fedora-devel-announce!)? A simple command line script is able to do that in case MTA's like sendmail or postfix are properly configured. That script likely could write a comma separated list that people can cut-n-paste to Thunderbird, Kmail and other MUA's if needed.
Fedora-devel-list is subscribed to fedora-devel-announce, so anything sent to f-d-a will also show up on f-d-l.
Reason: I for one slowly start to get annoyed by the slowly increasing traffic on fedora-devel-announce that clutters my inbox. Thus I started to consider to move mails from that list into some IMAP folder automatically, and that is exactly what we didn't want people to do when fedora-devel-announce was created. I think we said something like "less then 10 mails a month" back when we created it, but I could not find that on a quick google search :-/
Are you finding what is sent there to not be relevant or important to you as a packager? I'm the moderator of that list and I don't think I've sent or let anything through that wasn't useful.
Jesse Keating wrote:
Are you finding what is sent there to not be relevant or important to you as a packager? I'm the moderator of that list and I don't think I've sent or let anything through that wasn't useful.
Well, if the proposal is to send any and all soname bumps there, this will make it much higher volume. And really, nobody needs to know about "library libxyz which is only used by package xyz which is maintained by the same packager has a new soname". ;-)
Kevin Kofler
On Fri, 2009-03-06 at 23:42 +0100, Kevin Kofler wrote:
Jesse Keating wrote:
Are you finding what is sent there to not be relevant or important to you as a packager? I'm the moderator of that list and I don't think I've sent or let anything through that wasn't useful.
Well, if the proposal is to send any and all soname bumps there, this will make it much higher volume. And really, nobody needs to know about "library libxyz which is only used by package xyz which is maintained by the same packager has a new soname". ;-)
Kevin Kofler
Well sure, that's a tad ridiculous. However the solution there is to alter when its required to send, instead of creating yet another list to send the silly announcements to, or whatever else.
"JK" == Jesse Keating writes:
JK> On Fri, 2009-03-06 at 23:42 +0100, Kevin Kofler wrote:
Jesse Keating wrote:
Are you finding what is sent there to not be relevant or important to you as a packager? I'm the moderator of that list and I don't think I've sent or let anything through that wasn't useful.
Well, if the proposal is to send any and all soname bumps there, this will make it much higher volume. And really, nobody needs to know about "library libxyz which is only used by package xyz which is maintained by the same packager has a new soname". ;-)
Kevin Kofler
JK> Well sure, that's a tad ridiculous. However the solution there is to JK> alter when its required to send, instead of creating yet another list to JK> send the silly announcements to, or whatever else.
Indeed, which is why I also suggested in the thread above, the following exception in the response to Thorsten's original, which should address Kevin's concern:
However, there should definitely be a few exceptions to the rule of announcement if maintainer performing the breakage, either 1) owns all the affected packages and is planning to rebuild them, or 2) is a provenpackager/co-maintainer, or otherwise has access and who is intending to rebuild all affected packages (e.g. xulrunner/firefox). In those cases, a heads-up on f-d-l might be nice, but probably not mandatory (although even in thoses cases there may be dependent packages that the maintainer is not aware of).
Alex
On 06.03.2009 08:00, Jesse Keating wrote:
On Fri, 2009-03-06 at 07:03 +0100, Thorsten Leemhuis wrote:
Why not simply mail f-d-l as well as all the maintainers of packages that are affected (and not fedora-devel-announce!)? A simple command
^^^
line script is able to do that in case MTA's like sendmail or postfix are properly configured. That script likely could write a comma separated list that people can cut-n-paste to Thunderbird, Kmail and other MUA's if needed.
Fedora-devel-list is subscribed to fedora-devel-announce, so anything sent to f-d-a will also show up on f-d-l.
You misread; see the "not" above ;-)
Reason: I for one slowly start to get annoyed by the slowly increasing traffic on fedora-devel-announce that clutters my inbox. Thus I started to consider to move mails from that list into some IMAP folder automatically, and that is exactly what we didn't want people to do when fedora-devel-announce was created. I think we said something like "less then 10 mails a month" back when we created it, but I could not find that on a quick google search :-/
Are you finding what is sent there to not be relevant or important to you as a packager? I'm the moderator of that list and I don't think I've sent or let anything through that wasn't useful.
As Kevin replied: "If the proposal is to send any and all soname bumps there, this will make it much higher volume." That's what I fear, as the volume IMHO already is to high. Seems it wasn't obvious enough in my initial mail :-/ sorry.
To say it in different words: f-d-a afaik was created to make sure that all contributors get aware of all the *important* issues. That afaics will only work properly as long as the information that comes over the list most of the time is relevant for the one that receives it, otherwise people will consider the mails from the list as spam and filter them away -- which is exactly what we didn't want people to do when f-d-a was created.
That includes people that maintain just one single package in Fedora -- like for example grenier does, who takes care of testdisk upstream and in Fedora. Ask yourself: How many of the mails that went to f-d-a in February were relevant for developers like him -- developers that only want to maintain their own software in Fedora and nothing else (I guess some of those might not even use Fedora regularly....)
From a quick look on https://www.redhat.com/archives/fedora-devel-announce/2009-February/thread.h... I'd say: maybe half of the mails, likely way less.
The "Test Day" announcement for example. Or the soname bumps. FUDCon Berlin 2009. Announcing Fedora 11 Alpha (blink). None of them afaics is allowed according to the rules that are written on https://www.redhat.com/mailman/listinfo/fedora-devel-announce
Quoting: ---
This list is intended to be a LOW TRAFFIC announce-only list for Fedora development.
Acceptable Types of Announcements
- Policy or process changes that affect developers.
- Infrastructure changes that affect developers.
- Tools changes that affect developers.
- Schedule changes
- Freeze reminders
Unacceptable Types of Announcements
- Periodic automated reports (violates the INFREQUENT rule)
- Discussion
- Anything else not mentioned above
---
But it's likely a matter of interpretation.
Anyway: enough said. I think I made my point clear now and it's up to others to decide.
CU knurd
"TL" == Thorsten Leemhuis writes:
TL> On 05.03.2009 21:44, Jon Stanley wrote:
57 Make stronger policy/guideline that ABI/API/soname breakage should always be announced fedora-devel-announce (not f-d-l)
TL> Why not simply mail f-d-l as well as all the maintainers of packages TL> that are affected (and not fedora-devel-announce!)? A simple command TL> line script is able to do that in case MTA's like sendmail or postfix TL> are properly configured. That script likely could write a comma TL> separated list that people can cut-n-paste to Thunderbird, Kmail and TL> other MUA's if needed.
TL> Reason: I for one slowly start to get annoyed by the slowly increasing TL> traffic on fedora-devel-announce that clutters my inbox. Thus I TL> started to consider to move mails from that list into some IMAP folder TL> automatically, and that is exactly what we didn't want people to do TL> when fedora-devel-announce was created. I think we said something like TL> "less then 10 mails a month" back when we created it, but I could not TL> find that on a quick google search :-/
In a quick check on the archives, there seems to an average of around 15-20 e-mails per month on fedora-devel-announce (f-d-a) in recent months. Some months it is lower. I do not consider that high traffic.
Regarding sending to maintainers of packages that are affected, I agree that would also be a good idea, but it's often difficult to find out exactly which packages are affected. Yes, I know you can run repoquery to find those packages in most cases, but:
1) most maintainers are unaware of, or do not use repoquery regularly (even though they probably should)
To solve this, ideally we could perhaps create new aliases that maintainers could simply use to contact all downstream affected packages in one fell swoop, e.g.
<packagename>-dependent-packages-owners@fedoraproject.org
that would be populated by a list of e-mail addresses for owners of all dependent packages, which would be created by a repoquery run at initial package creation. It would also have to be updated as some kind of cron job as potentially dependent packages are added or removed from the repository.
2) even if repoquery is run does not necessarily collect all affected packages. In some cases these ABI changes have an effect beyond the immediate list of packages that repoquery will list. e.g. an update of the mono stack may affect a large number of packages and overall distro integration beyond what can be expressed by package deps.
Since a creation of such a system in (1) is probably a ways off and still doesn't necessarily solve (2), I still think mailing f-d-a makes the most sense. ABI/soname breakage are exactly the kind of thing that it is worth bringing to attention to the greater Fedora community, which is what f-d-a was intended to solve.
However, there should definitely be a few exceptions to the rule of announcement if maintainer performing the breakage, either 1) owns all the affected packages and is planning to rebuild them, or 2) is a provenpackager/co-maintainer, or otherwise has access and who is intending to rebuild all affected packages (e.g. xulrunner/firefox). In those cases, a heads-up on f-d-l might be nice, but probably not mandatory (although even in thoses cases there may be dependent packages that the maintainer is not aware of).
Lastly, all replies to f-d-a are redirected to f-d-l, and f-d-a is moderated, both of which should keep the list traffic to the currently manageable level.
Alex