Tag Archives: iss

Photos from Vilanova/Barcelona

Post Syndicated from Lennart Poettering original http://0pointer.net/blog/photos/barcelona.html

I finally found the time to sort my photos from Vilanova i la Geltrú and Barcelona.

My Windows of Barcelona series:

Windows of Barcelona

A few other nice shots:

Photo #361
Photo #371
Photo #366
Photo #381
Photo #386

Photo #222
Photo #210
Photo #125
Photo #137
Photo #5

Photo #311
Photo #301
Photo #317
Photo #281
Photo #269

Photo #268
Photo #89
Photo #49
Photo #35
Photo #95

These are:
1st row: Casa Milà; dito; dito; dito; dito;
2nd row: Palau de la Música Catalana; dito; Mies van der Rohe Pavilion; dito; Vilanova Lighthouse;
3rd row: Sagrada Família; dito; dito; Hospital de Sant Pau; dito;
4th row: Sagrada Família, seen from Sant Pau; City Center/Barri Gòtic; dito; dito; Plaça Reial

A panoramic view of Barcelona photographed from the Montjuic towards the north:

Barcelona Panorama

Those “thunderclouds” on the right side of the image are actually a
result of not using the same exposure settings on all photos that are
part of the panorama. Which is a mistake I didn’t repeat with my
second panoramic view, which again shows Barcelona from the Montjuic, but this time towards the east:

Barcelona Panorama 2

Dont miss the the entire album!

Polypaudio article on LWN

Post Syndicated from Lennart Poettering original http://0pointer.net/blog/projects/polypaudio-lwn.html

The current issue of the Linux
Weekly News
features a short article about Polypaudio. (The article is not (yet)
accessible for free, come back in a week if you aren’t an LWN subscriber.)

Quoting:

With its support for a wide variety of popular audio utilities, actively developed code, and broad capabilities, the Polypaudio project fills an important role in Linux-based audio development.

Polypaudio article on LWN

Post Syndicated from Lennart Poettering original http://0pointer.net/blog/projects/polypaudio-lwn.html

The current issue of the Linux
Weekly News
features a short article about Polypaudio. (The article is not (yet)
accessible for free, come back in a week if you aren’t an LWN subscriber.)

Quoting:

With its support for a wide variety of popular audio utilities, actively developed code, and broad capabilities, the Polypaudio project fills an important role in Linux-based audio development.

Avahi Gains Compatibility Layers for Apple Bonjour and HOWL

Post Syndicated from Lennart Poettering original http://0pointer.net/blog/projects/avahi-compat.html

A short while ago I checked in to SVN two API/ABI compatibility
modules which implement the HOWL and the Apple
Bonjour (dns_sd.h)
DNS-SD/mDNS APIs on top of Avahi’s
native API. Effectively this means that you can run *all*
Zeroconf-enabled software that is available for free operating systems
seamlessly on top of Avahi. Or at least the software that uses the
limited subset of API functions we support. Missing functions will be
implemented on an on-demand basis. Gnome-VFS/Nautilus works
perfectly, as does Gobby, which are the only real-world applications
we tested until now.

The list of supported/unsupported functions is available from SVN for HOWL and for
dns-sd.h.

The compatibility layers are actually pretty interesting pieces of code: for
compatibility with the way HOWL/Bonjour integrates with event loops we had to
hook up the timeout and I/O watches D-BUS depends on to a single file
descriptor. This involves all kinds of ugly things like threading and
“creative” ways to use the event loop abstraction Avahi provides. Some might
call this “cracktastic”, but it actually works pretty well.

The compatibility layers are not intended to be long term solutions. For
every session object we create a background thread that polls for events and a
DBUS session object. This is an utter waste of resources, especially on
dns_sd.h where every basic operation uses a session object of its own.
In addition, our compatibility layers are incomplete. We do not offer the full
set of functions or the full semantics. Our compatibility is just good enough
to make most Zeroconf-aware programs work with Avahi right now.

We consider neither dns_sd.h nor the HOWL API a “well designed”
API and encourage people to port their programs to our more powerful native
API. To stress this the two modules will warn the user about their usage and
write a warning line to STDERR and syslog. Hopefully this will annoy
people sufficiently that Avahi adoption speeds up a little.

To our own surprise we actually support at least one API function more than each of the
reference implementations! From dns_sd.h we support
DNSServiceEnumerateDomains() which is actually unsupported by
Apple Bonjour on POSIX/Linux systems. The documented HOWL function
sw_ipv4_address_decompose() is actually a NOOP in the
reference implementation, but isn’t in our compatibility layer.

Since dns_sd.h is the only file licensed under a BSD license in the otherwise APSL-licensed
mDNSResponder distribution, we were able to copy it into our sources untouched.

Here’s a screenshot of
Nautilus and Gobby
running on top of Avahi through the HOWL compatibility
layers.

Avahi Gains Compatibility Layers for Apple Bonjour and HOWL

Post Syndicated from Lennart Poettering original http://0pointer.net/blog/projects/avahi-compat.html

A short while ago I checked in to SVN two API/ABI compatibility
modules which implement the HOWL and the Apple
Bonjour (dns_sd.h)
DNS-SD/mDNS APIs on top of Avahi’s
native API. Effectively this means that you can run *all*
Zeroconf-enabled software that is available for free operating systems
seamlessly on top of Avahi. Or at least the software that uses the
limited subset of API functions we support. Missing functions will be
implemented on an on-demand basis. Gnome-VFS/Nautilus works
perfectly, as does Gobby, which are the only real-world applications
we tested until now.

The list of supported/unsupported functions is available from SVN for HOWL and for
dns-sd.h.

The compatibility layers are actually pretty interesting pieces of code: for
compatibility with the way HOWL/Bonjour integrates with event loops we had to
hook up the timeout and I/O watches D-BUS depends on to a single file
descriptor. This involves all kinds of ugly things like threading and
“creative” ways to use the event loop abstraction Avahi provides. Some might
call this “cracktastic”, but it actually works pretty well.

The compatibility layers are not intended to be long term solutions. For
every session object we create a background thread that polls for events and a
DBUS session object. This is an utter waste of resources, especially on
dns_sd.h where every basic operation uses a session object of its own.
In addition, our compatibility layers are incomplete. We do not offer the full
set of functions or the full semantics. Our compatibility is just good enough
to make most Zeroconf-aware programs work with Avahi right now.

We consider neither dns_sd.h nor the HOWL API a “well designed”
API and encourage people to port their programs to our more powerful native
API. To stress this the two modules will warn the user about their usage and
write a warning line to STDERR and syslog. Hopefully this will annoy
people sufficiently that Avahi adoption speeds up a little.

To our own surprise we actually support at least one API function more than each of the
reference implementations! From dns_sd.h we support
DNSServiceEnumerateDomains() which is actually unsupported by
Apple Bonjour on POSIX/Linux systems. The documented HOWL function
sw_ipv4_address_decompose() is actually a NOOP in the
reference implementation, but isn’t in our compatibility layer.

Since dns_sd.h is the only file licensed under a BSD license in the otherwise APSL-licensed
mDNSResponder distribution, we were able to copy it into our sources untouched.

Here’s a screenshot of
Nautilus and Gobby
running on top of Avahi through the HOWL compatibility
layers.

Avahi on Linux Weekly News

Post Syndicated from Lennart Poettering original http://0pointer.net/blog/projects/avahi-lwn.html

Seems today’s edition of LWN features a front page story about Avahi. It’s actually quite nice, even though I missed an emphasis on the fact that Avahi’s mDNS stack itself is embeddable into applications via a shared library.

I guess you’ll have to wait a week if you want to read the article without subscription.

Avahi on Linux Weekly News

Post Syndicated from Lennart Poettering original http://0pointer.net/blog/projects/avahi-lwn.html

Seems today’s edition of LWN features a front page story about Avahi. It’s actually quite nice, even though I missed an emphasis on the fact that Avahi’s mDNS stack itself is embeddable into applications via a shared library.

I guess you’ll have to wait a week if you want to read the article without subscription.

Avahi 0.1 Looming

Post Syndicated from Lennart Poettering original http://0pointer.net/blog/projects/avahi-0.1.html

Avahi 0.1 is due in
the next few days. The last missing piece is a simplifying C wrapper around the
DBUS API. Though Avahi is currently pre-0.1 it is already quite complete and
mature. To put it with Ross Burton: “… this doesnt count as 0.1 because it
has docs, man pages *and* works”

Unfortunately python-dbus has quite a few bugs which make it very difficult
to code with. e.g. it doesn’t handle sending empty arrays, fails to send byte
values and so on. It is difficult to work around all these issues, therefore
the Avahi client tools will not work with an unpatched python-dbus. You need to
apply this
patch
(applying to 0.35.2) to fix at least the byte value bug to get
them working.

Avahi 0.1 Looming

Post Syndicated from Lennart Poettering original http://0pointer.net/blog/projects/avahi-0.1.html

Avahi 0.1 is due in
the next few days. The last missing piece is a simplifying C wrapper around the
DBUS API. Though Avahi is currently pre-0.1 it is already quite complete and
mature. To put it with Ross Burton: “… this doesnt count as 0.1 because it
has docs, man pages *and* works

Unfortunately python-dbus has quite a few bugs which make it very difficult
to code with. e.g. it doesn’t handle sending empty arrays, fails to send byte
values and so on. It is difficult to work around all these issues, therefore
the Avahi client tools will not work with an unpatched python-dbus. You need to
apply this
patch
(applying to 0.35.2) to fix at least the byte value bug to get
them working.

The GNU GPL and the American Dream

Post Syndicated from Bradley M. Kuhn original http://ebb.org/bkuhn/blog/2001/02/21/american-dream.html

[ This essay
was originally
published on gnu.org
. ]

When I was in grade school, right here in the United States of America,
I was taught that our country was the “land of opportunity”. My teachers
told me that my country was special, because anyone with a good idea and a
drive to do good work could make a living, and be successful too. They
called it the “American Dream”.

What was the cornerstone to the “American Dream”? It was
equality — everyone had the same chance in our society to choose
their own way. I could have any career I wanted, and if I worked hard, I
would be successful.

It turned out that I had some talent for working with computers —
in particular, computer software. Indoctrinated with the “American
Dream”, I learned as much as I could about computer software. I
wanted my chance at success.

I quickly discovered though, that in many cases, not all the players in
the field of computer software were equal. By the time I entered the
field, large companies like Microsoft tended to control much of the
technology. And, that technology was available to me under licensing
agreements that forbid me to study and learn from it. I was completely
prohibited from viewing the program source code of the software.

I found out, too, that those with lots of money could negotiate
different licenses. If they paid enough, they could get permission to
study and learn from the source code. Typically, such licenses cost many
thousands of dollars, and being young and relatively poor, I was out of
luck.

After spending my early years in the software business a bit
downtrodden by my inability to learn more, I eventually discovered another
body of software that did allow me to study and learn. This software was
released under a license called the GNU General Public License (GNU
GPL). Instead of restricting my freedom to study and learn from it, this
license was specifically designed to allow me to learn. The license
ensured that no matter what happened to the public versions of the
software, I’d always be able to study its source code.

I quickly built my career around this software. I got lots of work
configuring, installing, administering, and teaching about that
software. Thanks to the GNU GPL, I always knew that I could stay
competitive in my business, because I would always be able to learn easily
about new innovations as soon as they were made. This gave me a unique
ability to innovate myself. I could innovate quickly, and impress my
employers. I was even able to start my own consulting business. My own
business! The pinnacle of the American Dream!

Thus, I was quite surprised last week
when Jim Allchin, a
vice president at
Microsoft hinted
that
the
GNU GPL
contradicted
the
American Way.

The GNU GPL is specifically designed to make sure that all
technological innovators, programmers, and software users are given equal
footing. Each high school student, independent contractor, small business,
and large corporation are given an equal chance to innovate. We all start
the race from the same point. Those people with deep understanding of the
software and an ability to make it work well for others are most likely to
succeed, and they do succeed.

That is exactly what the American Way is about, at least the way I
learned it in grade school. I hope that we won’t let Microsoft and
others change the definition.

Finished Thesis

Post Syndicated from Bradley M. Kuhn original http://ebb.org/bkuhn/blog/2001/01/22/masters-complete.html

My thesis is nearly complete. I defend tomorrow, and as usual, I let
the deadline run up until the end. I just finished my slides for the
defense, and practiced once. I have some time in the schedule tomorrow to
practice at least once, although I have to find some empty room up at the
University to do it in.

I’ll be glad to be done. It’s been annoying to spend three or four
weeks here sitting around writing about perljvm, and not hacking on it. I
have a Cosource deadline coming up this week, so now’s a good a time as
any to release the first version of the Kawa-based perljvm.

I am really excited about how Kawa works, and how easy it is to massage
perl’s IR into Kawa’s IR. I got more excited about it as I wrote my thesis
defense talk. I really think great things can happen with Kawa in the
future.

Larry Wall is here, and we’ve had two dinners for the Cincinnati
GNU/Linux Users’ Group (who paid Larry’s way to come here). I was there,
and Larry was asking some hard-ish questions about Kawa. Not hard exactly,
just things I didn’t know. I began to realize how much I have focused on
the Kawa API, and I haven’t really been digging in the internals. I told
him I’d try to have some answers about it for my defense, and I will
likely reread Bothner’s papers on the subject tomorrow to get familiar
with how he deals with various issues.

It’s odd having Larry on my thesis committee. I otherwise wouldn’t be
nervous in the least, but I am quite worried with him on the
committee.

Anyway, so I defend tommorrow, then it’s into perljvm hacking again
right away on Tuesday to make the Cosource deadline, and then I have to
finish preparing my Perl tutorial for LinuxExpo Paris.