I do not know much in these areas having learned my craft by digging
UNIX source code listings out of wastebaskets at Harvard (freshman
weren't really supposed to have access to the UNIX source code in
1974). And I hardly ever used the Oracle bug reporting system in the
mid-1980s. But I have used the Mozilla, Gnome and Gentoo public bug
reporting and tracking systems to a fair extent in recent years and
they seem to work fairly effectively.
They have good search engines. They attach patches to bug reports
which fix the stated problems. They allow for discussion, assessment,
voting and presumably for the powers in charge the keeping of
statistics regarding the state of the software.
Now, the LKML seems to to be a throwback almost to 1971 when the first
email messages were sent between a couple of PDP-10s. It has patches
that I have no interest in, discussions I have no interest in and were
it not for Gmail's search function it would be generally useless (very
high Noise-to-Signal ratio). Now maybe I do not understand the Linux
development process. Maybe this is a "Wizard of Oz" case and there is
a hidden bug reporting system hidden behind the curtain -- but in
spite of my best efforts I cannot locate it.
Why in this day and age (ignoring historical inertia) has Linux failed
to adopt a robust (modern) bug reporting system?
Can this be fixed? Or is Linux really a case of a "proprietary
software" system pretending to be open [1]?
Robert Bradbury
1. If a software system is so complex that its quirks and pitfalls
cannot easily be located and avoided (witness the ondemand scheduler
problem on Pentium IV's message I recently filed) then is it not
*effectively* open source. I am qualified to read hardware manuals, I
am qualified to rewrite C code (having written code generators for
several C compilers) but the LKML is like the windmill and I feel like
Don Quixote tilting back and forth in front of it. One could even
argue that the lack of an open bug reporting system (and "current
state" online reports) effectively makes Linux a non-open-source
system. Should not Linux be the one of the first systems to make all
knowledge completely available? Or is it doomed to be replaced by
systems which might provide such capabilities (Android perhaps???)
Robert Bradbury wrote:
> Now, the LKML seems to to be a throwback almost to 1971 when the first
> email messages were sent between a couple of PDP-10s. It has patches
> that I have no interest in, discussions I have no interest in and were
> it not for Gmail's search function it would be generally useless (very
> high Noise-to-Signal ratio). Now maybe I do not understand the Linux
> development process. Maybe this is a "Wizard of Oz" case and there is
> a hidden bug reporting system hidden behind the curtain -- but in
> spite of my best efforts I cannot locate it.
You didn't read http://www.kernel.org/ ??
There is a nice section on the first page stating:
Reporting Linux Kernel bugs
Please see http://www.kernel.org/pub/linux/docs/lkml/reporting-bugs.html
if you want to report a Linux kernel bug. Bug reports sent to the
kernel.org administrators will be ignored.
There is now a bugzilla setup at bugzilla.kernel.org. Currently this is
for reporting kernel version 2.6 bugs only.
--
Thomas
Thomas, thank you for pointing this out.
I was unaware of the kernel.org bugzilla database. Sometimes people
will go from AT&T UNIX to SGI IRIX to Linux without making a stop at
kernel.org. If it comes on a CD and boots relatively easily people
may not bother with such details. I think I started using Linux
around the time I started using Mosaic and that may have been before
the kernel.org site was as robust as it now appears to be (if it
existed at all).
Given the bugzilla database, is the LKML with its high volume and
noise level (IMO), now largely obsolete [1]? (I think I a Google
search may have led me in the direction that it was the place to
report bugs).
1. Is there an lkml-interesting sublist (or do subscribers need to
make up their own filters)?
On 10/25/09, Thomas Backlund <[email protected]> wrote:
> Robert Bradbury wrote:
>> Now, the LKML seems to to be a throwback almost to 1971 when the first
>> email messages were sent between a couple of PDP-10s. It has patches
>> that I have no interest in, discussions I have no interest in and were
>> it not for Gmail's search function it would be generally useless (very
>> high Noise-to-Signal ratio). Now maybe I do not understand the Linux
>> development process. Maybe this is a "Wizard of Oz" case and there is
>> a hidden bug reporting system hidden behind the curtain -- but in
>> spite of my best efforts I cannot locate it.
>
> You didn't read http://www.kernel.org/ ??
>
> There is a nice section on the first page stating:
>
>
> Reporting Linux Kernel bugs
>
> Please see http://www.kernel.org/pub/linux/docs/lkml/reporting-bugs.html
> if you want to report a Linux kernel bug. Bug reports sent to the
> kernel.org administrators will be ignored.
>
> There is now a bugzilla setup at bugzilla.kernel.org. Currently this is
> for reporting kernel version 2.6 bugs only.
>
>
> --
> Thomas
>
Robert Bradbury wrote:
> Given the bugzilla database, is the LKML with its high volume and
> noise level (IMO), now largely obsolete [1]?
No, it's not. Many issues benefit from a mail-based discussion where many
people can follow the thread and jump in if something strikes them.
People subscribed to lkml will generally have their own methods of
filtering out mails/subjects/threads that interest them.
Some subsystems (like acpi) prefer to handle bugs using bugzilla; others
(netdev, ide) have dedicated mailing lists. All regressions are tracked in
bugzilla, even if many are initially reported on mailing lists.
lkml still has an important function for those who wish to keep a good
general overview of what's happening. There are a few people who keep an
active watch on lkml to spot bug reports and to ensure they are brought to
the attention of the right people.
You are not required to subscribe to lkml to report an issue on it: the
policy is to always reply-to-all.
If you prefer to use bugzilla, go ahead. Some subsystem maintainers will
then handle the bug in bugzilla; others may switch to email.
> (I think I a Google search may have led me in the direction that it was
> the place to report bugs).
It depends. It's the one place where you can always report an issue if you
don't know where else to report it. The best way to report issues is to do
so directly to relevant maintainers, but that requires some familiarity
with the source and the community.
Cheers,
FJP
On Mon, Oct 26, 2009 at 08:12:23AM -0400, Robert Bradbury wrote:
> I was unaware of the kernel.org bugzilla database. Sometimes people
> will go from AT&T UNIX to SGI IRIX to Linux without making a stop at
> kernel.org. If it comes on a CD and boots relatively easily people
> may not bother with such details. I think I started using Linux
> around the time I started using Mosaic and that may have been before
> the kernel.org site was as robust as it now appears to be (if it
> existed at all).
>
> Given the bugzilla database, is the LKML with its high volume and
> noise level (IMO), now largely obsolete [1]? (I think I a Google
> search may have led me in the direction that it was the place to
> report bugs).
Robert,
Most end users who use a Linux distribution will have a much better
time if they use the support channels (bugzilla, irc, web forums,
etc.) for their particular Linux distribution. So if you are using
Ubuntu, use the Ubuntu bugzilla and web forums. If you are using
Fedora, they have their own bugzilla and web forums. Debian users
have the Debian BTS (Bug Tracking System), as well as many mailing
lists (focused on debian newbies, debian developers, etc.)
The LKML and the kernel bugzilla are best used for people who are
working on the mainline Linux kernel, as opposed to the kernels which
have been snapshotted and customized for each distribution. Community
distributions, which tend to snapshot every six months, do tend to be
closer to mainline than Enterprise kernels (RHEL, SLES) that snapshot
every 24 months or so, but even some community distributions, such as
Ubuntu, do add a significant number of "value add" (and thus, bugs :-)
to their kernels.
There are also individual sublists for people who are only working on
a specific part of the kernel (i.e., the linux-ext4 mailing list, the
linux-scsi mailing list, etc).
- Ted
Theodore Tso wrote:
> Most end users who use a Linux distribution will have a much better
> time if they use the support channels (bugzilla, irc, web forums,
> etc.) for their particular Linux distribution.
[...]
> even some community distributions, such as
> Ubuntu, do add a significant number of "value add" (and thus, bugs :-)
> to their kernels.
The downside is that such bugtrackers appear to be chronically
understaffed in the triage department. Therefore, bugs which could look
like potential upstream bugs to a somewhat trained eye are far too
rarely reported to upstream by distro bug triage teams.
> There are also individual sublists for people who are only working on
> a specific part of the kernel (i.e., the linux-ext4 mailing list, the
> linux-scsi mailing list, etc).
Robert,
these mailing lists can be found in the MAINTAINERS file in the root of
the kernel sources. The reporting-bugs page which Thomas mentioned
points to this file as a source for contact persons, but the list
addresses in there are IMO even more important.
About bugzilla.kernel.org: It's more a tracker than a reporting tool.
--
Stefan Richter
-=====-==--= =-=- ==-=-
http://arcgraph.de/sr/
> Given the bugzilla database, is the LKML with its high volume and
> noise level (IMO), now largely obsolete [1]? (I think I a Google
> search may have led me in the direction that it was the place to
> report bugs).
>
> 1. Is there an lkml-interesting sublist (or do subscribers need to
> make up their own filters)?
There's (or was?) lkml summary podcast ...
Pavel
--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html