In the Advogato discussion "Quality vs Quantity" (Article #157)
mettw suggested the idea of a journal where free
software developers could submit articles about their projects in a
scientific fashion. As a first step towards this I've created a "free
software tech-reports" announcement list (firstname.lastname@example.org) for
posting announcements of technical reports, papers and other types of
formal publication about free software projects.
This will work in a
similar way to the "preprint servers" used in the scientific
community: authors post the title and abstract of their work to the
list in a standard format, and subscribers receive an email digest of
new announcements. The full text of each paper is accessible in
postscript or pdf via a url in the message.
The list is going to cover any software under GPL-compatible
licenses -- a very wide scope. The aim is to have something which
will alert people to articles they might not otherwise have found, and
give everyone an overview of the major developments in other projects.
I will moderate the list to keep up a professional standard and
generally try to make it unobtrusive.
To subscribe to the announcement list, send email to
with "subscribe" in the subject or body.
If you've written an article recently, or in the past, please go
ahead and submit an announcement. Articles need to be freely
redistributable, either under a "copying permitted provided this
notice is preserved"-type license or a free documentation license such
as the GNU FDL. This is so that people can distribute the articles
along with the associated software. If you have any questions about
the procedure there is a email@example.com mailing list for
general queries. I will hold the first postings in the moderation
queue until a sufficient number of subscribers have had a chance to
For more details see the
firstname.lastname@example.org page at mail.gnu.org.
Yes, when journals require a transfer of copyright they usually
want to prevent distribution of the article by anyone else,
including the authors.
Their economic model has evolved in the
pre-digital age --- they own and control
access to information, rather than providing a service of redistributing it. This is the same situation as proprietary
software companies vs free software redistributors. Technological
change means that these journals now restrict the flow of information, compared with what is possible.
The solution to the problem is to use journals which don't
restrict redistribution. For example, in AI there is the Journal of Artificial
Intelligence (JAIR). If there isn't a suitable free journal then
one can be created by people in the field. Ultimately this is the only way that their work can be made available to everyone. I have suggested this on the list's
page -- I think it would be good to have a freely redistributable
journal like "Software Practice and Experience" but for free software.
There is a email@example.com list for further ideas.
Last May I decided to test the 2.4.0-test1 kernel on my slackware laptop and soon
found a bug in which my Adaptec APA-1480 Cardbus SCSI Host Bus Adapter
I looked in the kernel sources and found instructions about what to do
when one had this kind of bug - report it to the linux-kernel mailing list.
Well, that's what I did and although linux-kernel accepts messages from
unsubscribed users (so they get lots of headhunter spam), I wanted to
follow up to be sure the bug got fixed so I subscribed for a while. And
boy, didn't I get a lot of email.
Further, it took some work to get the bug fixed. People would post
patches, and I'd reply with notifications that the patches worked, or
maybe that there were additional problems, but it took some effort to
see that the bug fixes stuck.
I was OK with this being a mailing list junkie and a developer, but I
felt that this wouldn't really work for the user at large, and probably
in reality wasn't all that great for the kernel developers either.
So I made a suggestion entitled Organized
Linux QA? in which I suggested:
There are a lot more people who are capable and willing to simply try
out configuring and building a new kernel and to report the results than
there are who can track down the cause of the bugs, but a lot more hours
are needed just to try out that huge permutations of configurations
needed to find that a bug exists.
What I was thinking would be useful would be to have a database where
people who wanted to volunteer to do QA could register themselves, and
then enter the configurations of machines that they have available to
them - brand of machine, CPU time, PCI and ISA cards, and so on. They
could have URLs for pages with further descriptions of the machine.
Each user would give a name to a preset configuration and then when
reporting bugs would indicate what preset they'd used (I'd longed for
this at companies I'd worked at).
There would be a contact method there. Either the user could log in and
check their messages, or indicate that messages meant for them should be
dropped in their email.
Then a kernel developer (and it could be expanded to include any free
software) could send a message to anyone who had a PhooTek PCI
WhizzoMaster Card and say "I've got a new driver available, please try
it out", and then everyone would try it out and be able to file not just
bugs, but successful tests.
Testers could log relevant information such as the .config files they'd
used for particular tests and so on.
Further, you could do boolean searches such as "show me all the test
results with machines that have a 3C-905 ethernet card and an adaptec
29160 SCSI host bus adapter with CONFIG_HAPPYMEAL set in their kernel
As a step towards making this easy for people to find, I registered the
domains "linuxquality.org" and .com. A number of folks offered to host
the site, my feeling for the best choice was a certain server facility
that enthusiastically supports open source and that has a fat net
I got a generally positive response to this suggestion, although it was
suggested that folks like Linus and Alan Cox would need to
One thing that's important to understand is that the Linux kernel
development is so distributed that not everyone who contributes is
actually connected to the Internet. Some folks mail their patches in
via gateways and even on floppy disk (or so I've heard) and so a web
database would cut these folks off if that was the only way they could
get their bugs. One solution would be to allow developers to request
that summaries be emailed to them, or even hardcopies sent regularly via
Unfortunately, my consulting
business got real busy with the need to ship a product and I was
also married in July. So I never got time to even start on this. I'm
not out of the woods yet but there is hope that I will have a normal
life again soon, and I'd like to pursue this. I just sent an email to
the folks who'd offerred to host it asking if they're still