It sounds like your proposal was a lot smoother than my own. We were in Rome over Easter, and our hotel (in the old part of the city) had a wonderful view over the city from the roof terrace. I was forming grand plans of proposing up on the terrace, with the city lights and fine wine... Then it turns a bit to cold to sit outside at 11pm and our hotel didn't sell wine, only small cans of beer. Not quite the vision I had, but a million things could have ruined the vision (someone else on the terrace for example). However, everything from that point went to plan and I still find myself grinning uncontrollably when I realise that in under 3 months we'll be Mr and Mrs!

I mean, people. With this code:

int foo(void) {
  const int i = 42;
  return i + 3;
}

What compiler doesn't just optimise the body away to "return 45" when I ask for optimised code? Well, one which also doesn't optimise away the test in do {...} while(0), that is the answer.

This compiler has a positive plethora of optimisation options: you can optimise for either speed or size, from either level 0 to level 9. Set to size optimisation level 9, the generated code goes something like:

Grow stack by 2 bytes
Put 42 in R16
Put 0 in R17
Store R16 via Z (stack pointer)
Store R17 via Z+
Put 45 in R16
Shrink stack by 2 bytes
Return (R16/R17 pair is the return value)

Nothing quite like 6 wasted instructions out of 7 on a <4MHz processor... When people moan about GCC not doing the right thing, they don't know the meaning of "broken compiler". Amusingly the compiler didn't generate the code to clear R17 to 0, so it either noticed that had already happened, or the code generator is even worse than I thought. I'm not sure what case to believe in.

NP: When It Falls, Zero 7.

• Port to GStreamer 0.10

Thanks to Tim-Philipp Mueller and James Livingston for helping me here. This needs lots of testing, there are bound to be many regressions that need hunting and fixing. Everyone grab it now!

• Bind F1 to Help
• [update] Kick-arse updated documentation (thanks Shaun!)

But plenty of changes for people who haven't used 2.11 including CD playback, threaded extraction, uses gnome-vfs to write the songs, and genre support.

Thanks to the ever-working translators: Danilo Šegan (sr), Mohammad Damt (id), Clytie Siddall (vi), Jean-Michel Ardantz (fr), Michiel Sikkes (nl), Roozbeh Pournader (fa).

• Handle errors when transforming filename encoding fails (Frederic Crozat)
• Handle MusicBrainz saying it has found a matching album when it hasn't (FC)
• Fix the "time remaining" calculations (hondaguru)
• Updated libbacon, fixing various issues
• Fix crashes due to the idle handler not being removed (Colin Walters)
• Stop using the deprecated _() in libgnome (Mariano Suárez-Alvarez)

NP: Two Pages, 4 Hero

• Don't use wnck_window_set_geometry as it was introduced in GNOME 2.16 (#381233).

Downloads are in the usual place.

This is something I totally agree with, whilst FlickrAuth works and was clearly an important influence on OAuth, it's a single-service protocol when OAuth has managed to get massive adoption and a huge developer base.

The problem I've got is libsocialweb, which has a Flickr module that allows both fetching of your contact's recent photos and uploading images. This uses FlickrAuth so at the end of July will suddenly stop working. I've got enough on my plate at the moment and would love for more people to understand how the entire social spagetti works, so this is a call for a volunteer to work on this, for which I'll obviously be available to offer any guidance and mentoring required.

There are two ways of approaching this, the easy way and the slightly harder way.

The easy way is to update libsocialweb to use an OAuthProxy instead of a FlickrProxy, and update the module metadata so that Bisho uses the generic OAuth flow instead of a Flickr-specific flow. This should be fairly simple and needs to happen soon so that any distributions that are using libsocialweb don't break in the summer.

The harder way is to add Flickr support to gnome-online-accounts, using the Twitter service as an example, and then port the Flickr service in libsocialweb to use gnome-online-accounts to authenticate. I've a proof of concept for the librest-goa integration which will be a useful starting point. This is more of a proof of concept for libsocialweb, we've been looking at moving away from Bisho but haven't done anything substantial yet.

Ideally both of these happen, so the current code will continue to work in the future and the GOA work demonstrates how GOA and libsocialweb would work together. So, anyone interested?

One day I'll find the time to integrate this properly into the network menu...

$ ./iphonemon.py 
Found Ross Burton’s iPhone
3 3G
1 3_75G
2 3_75G
3 3_75G
2 3_75G
3 3_75G
2 3G

Next step: a visual interface.

GSettings, like GConf before it, allows the administrator of a system to override the default settings or lock down keys to particular settings. This is well documented in the GNOME wiki.

However GConf didn't really have the concept of vendor patches. Traditionally if a Vendor wanted to change a default (say, the wallpaper) they'd have to patch the GConf schemas directly. Luckily for people who maintain distributions, GSettings provides a way of installing vendor overrides directly. It's not documented as far as I can tell so consider this a first draft at the manual...

First, find the setting you want to override, dconf-editor is useful for this. Say you're making a work-orientated custom distribution so you want the shell's popup calendar to show the week number by default. Some digging in dconf-editor leads us to the org.gnome.shell.calendar folder (the "schema") with a boolean key show-weekdate that defaults to false. By changing the default of this to true, all users will have work week shown unless they explicitly set it to false.

Now we've found the information we need we can write the override file. Create a new file with the extension .gschema.override, such as mydistro-tweaks.gschema.override. This file is a .ini-style keyfile, logically mapping schemas to groups and key/value pairs to (predicable) key/value pairs. The value needs to be in the GVariant serialisation format, but for things like booleans, numbers and strings these are fairly obvious. So we'd have a file that looks a little something like this:

[org.gnome.shell.calendar]
  show-weekdate=true

Note that you can set multiple keys in multiple schemas in the same override file, so if we also wanted to show the date in the panel we'd have this:

[org.gnome.shell.calendar]
show-weekdate=true

[org.gnome.shell.clock]
show-date=true

Now the file is ready to be installed. Put it in a package, install to $prefix/share/glib-2.0/schemas and finally run glib-compile-schemas $prefix/share/glib-2.0/schemas in the post-install/post-remove hooks. Done!

(many thanks to Ryan Lortie for telling me how vendor patches work)

I'd actually like this so much that I'm willing to put up some of my own hard cash to see it happen. Are there sites that will let people pledge money towards projects like this?

NP: Central Reservation, Beth Orton

• Lots of translations
• Fix i18n in the about dialog

Yeah, it's all go on the Tasks front... Tarballs on the Pimlico site, or gnome.org.

NP: Yanqui U.X.O. - Godspeed You! Black Emperor

Born at home (planned) on Thursday 4th November 2010 at 19:08, weighing 6 pounds 14.5 ounces. Both mother and baby are well. I'm now on leave until the end of December, so expect delayed responses to my personal email and vacation autoreplies to my Intel email.

flickrest

A Python/Twisted library for the Flickr API. This was written for use in Postr, although I suspect now that I don't maintain Postr any more they have forked. Maybe now this is in Git we can merge any changes.

evo-known-contact

A small tool I wrote for someone years ago that takes an RFC2822-formatted email on stdin, extracts the sender, and sets the exit code depending on whether that email address is in the address book.

feednotify

Display notifications when a RSS feed is updated.

Zebu

A tool to manage Debian chroots using cowbuilder.

Tumblrss

Screen-scrape your Tumblr dashboard and generate a RSS feed. This has bitrotted but was very useful.

gupnp-scrobbler

Listen to announcements over UPnP of music being played and submit the tracks to Last.fm.

ephy-gupnp

Dynamically generate bookmarks from UPnP devices that expose a web interface. Probably doesn't work with recent Epiphany releases because I've switched to Chrome.

ephydeli

An action to add the current page to Delicious.com. Probably doesn't work with recent Epiphany releases because I've switched to Chrome and Pinboard.

eds-tools

Some tools I wrote when working on EDS such as a dummy addressbook backend and command-line access to the libebook API.

cdscrobbler

Submit the current CD (or an arbitrary MusicBrainz album ID) to Last.fm as if you'd just finished playing it.

Out of all of these hacks I only actively use flickrest and Zebu now, so I wouldn't be surprised if there is some serious bitrot in the others. Hopefully something here is useful to someone, somewhere though!

• Huge translation update, including several missing strings
• Add a "tomorrow" button to the date popup
• Support adding tasks from the command line
• Use "category" over "group" consistantly
• Ensure the entry is correctly styled
• Ellipzies categories in the combo box
• Correctly encode non-ASCII notes
• Fix compilation with GTK+ 2.18

Tarballs and more information as usual are available at the Pimlico Project web site.

In related news, we're slowly migrating over to the GNOME infrastructure. We've migrated the source code, next up is the tarballs and bugzilla.

• Support the Nokia N810 integrated GPS. If someone can verify that this works for the N900 too, that would be great.
• Ability to dump the parsed NMEA to the console for debugging
• Fixed over-eager old-school Garmin detection
• Support reading NMEA from named pipes and FIFOs
• Support seting the baud rate on ghetto GPS devices that don't default to a baud rate that actually works (Globalsat ND-100 and BlueNext BN-903S, I'm looking at you)
• Support NMEA < 2.3

Many thanks to Jussi Kukkonen for patch review, and Bastien Nocera for patch review and new features.

The big question of course is what of the future? So far we've got some rough ideas. An overhaul of the device interaction layer is definitely required as actaully getting NMEA is becoming more complex: for integrated 3G/GPS chips you need to talk to oFono/ModemManager to get a socket, for some embedded GPS devices you need a proprietary binary that writes to a pipe, and so on. There are some new features we're considering too: server-side proximity detection and update rate limiting.

• Many translations
• Use gnome-session instead of gnome-power-manager to avoid the machine going to sleep (Richard Hughes)
• Fix a few crashers when extracting an unknown CD (Bastien Nocera)
• Fix CD-Text metadata gathering (BN)
• Don't truncate submission URLs (BN)
• Extract UUIDs to put in ripped files' metadata (Philipp Wolfer)
• Fix some bugs in test program (Alex Larsson)

Bastien originally called this release Not the maintainer, lalala, plug ears but we all know he is, right?

Now all I need is for someone to adopt Sound Juicer...

A trip to London on Saturday, in practise. Leave Ely 16:26, arrive Kings Cross 10 minutes late. Change to Piccadilly line and stand outside the closed barriers for 15 minutes because of overcrowding. Give up on the tube, catch a number 59 bus to Aldwych: 10 minutes to reach Euston (faster to walk) and gave up after sitting in gridlock on Russel Square for 15 minutes. Eventually get rather empty Piccadilly line from Russel Square to Covent Garden, arriving 18:50. Oh, and then the restaurant said it would be a two hour wait for a table for four.

That said, the return journey wasn't exactly a barrel of laughs. The tube was behaving so that took the expected ten minutes, but then we just missed the train back home and the next train wasn't for over an hour. Jump into a taxi to Liverpool Street to catch the 22:26... to discover there are engineering works and we'd have to get a bus for two hours. Attempt to get the tube back to Kings Cross... more engineering works so that was out. Another taxi back to Kings Cross and we finally get on the last train home, arriving at 00:35.

Just for extra fun I'm in London for the Moblin 2.0 Release Party on Monday and there are yet more engineering works, so if I miss the 22:15 I'll be on a bus for half the journey. Stab stab stab.

• Many translations
• Updated documentation
• Disable paranoia on playback (Bastien Nocera)
• Fix leaks and crashes in the metadata fetches (BN)

Did I mention that SJ could really do with a dedicated (co)maintainer?

NP: Cold Water Music, AiM

Now to finish reviewing the calendar port and merge that too...

NP: Session 2 - The Herbaliser Band

Packages for Debian are in the upload queue now, and I believe everyones favourite frockney is working on updating Fedora now.

NP: Oneric, Boxcutter

Because the world is rapidly moving to OAuth 1.0a exclusively after a rather painful attack was discovered against 1.0, I've recently been updating our bling HTTP/REST/XML IPC library librest to support it. In particular Twitter only supports 1.0a, and Fire Eagle shows the user a very scary message unless 1.0a is used. Now that the code is finished I thought I'd give a example of the new API when used with Twitter.

#include <rest/oauth-proxy.h>

Including the OAuthProxy headers is a good start.

int
main (int argc, char **argv)
{
  GError *error = NULL;
  RestProxy *proxy;

  g_thread_init (NULL);
  g_type_init ();

  proxy = oauth_proxy_new ("UfXFxDbUjk41scg0kmkFwA",
                           "pYQlfI2ZQ1zVK0f01dnfhFTWzizBGDnhNJIw6xwto",
                           "https://twitter.com/", FALSE);

First, initalise the GLib threading and type system. Threading is required by libsoup at the moment because it will use threads to lookup names in the background, I imagine this requirement will disappear with the next GLib release.

Next, an OAuthProxy is created. The two strings of garbage are our OAuth Consumer Key and Consumer Secret, then the URL endpoint to access and FALSE to say that this URL is complete and doesn't require expansion. Yes, that was Consumer Secret. Not very secret, is it.

  if (!oauth_proxy_request_token (OAUTH_PROXY (proxy), "oauth/request_token", "oob", &error))
    g_error ("Cannot get request token: %s", error->message);

Here we ask for a Request Token. The function to call is oauth/request_token, and because this is a basic test application which doesn't support URI callbacks we're setting the callback URI to oob (out-of-band). It is the callback URI argument that tells the server that we're using OAuth 1.0a, in 1.0 this parameter (oauth_callback at the HTTP leve) doesn't exist.

The callback is used to pass from the server to the client a verifier which is then required to obtain the Access Token. In the case of Twitter, this is a seven digit number. If a URI was specified then it would be invoked with the verifier as a query argument, but because we're getting it out-of-band Twitter will show it to the user and ask them to enter it into the application.

  g_print ("Go to http://twitter.com/oauth/authorize?oauth_token=%s then enter the PIN\n",
           oauth_proxy_get_token (OAUTH_PROXY (proxy)));
  fgets (pin, sizeof (pin), stdin);
  g_strchomp (pin);

Here we tell the user to go to the authorisation URL (to which we add the request token we have so far), and then enter the PIN that Twitter gives them.

  if (!oauth_proxy_access_token (OAUTH_PROXY (proxy), "oauth/access_token", pin, &error))
    g_error ("Cannot get access token: %s", error->message);

Now we ask for an Accesss Token. The function to call is oauth/access_token, and we're passing the PIN the user entered as the validator. If we were using OAuth 1.0 then the validator would be NULL.

If this method succeeds then we have an Access Token, and are authenticated. To avoid the authentication dance the Access Token and Token Secret should be saved somewhere secure (gnome-keyring would be a good idea) for future use.

  RestProxyCall *call;
  call = rest_proxy_new_call (proxy);
  rest_proxy_call_set_function (call, "statuses/update.xml");
  rest_proxy_call_set_method (call, "POST");
  rest_proxy_call_add_param (call, "status", "Hello from librest!");
  if (!rest_proxy_call_sync (call, &error))
    g_error ("Cannot make call: %s", error->message);
  return 0;
}

First a Call object is created, which encapsulates all of the data required to make a REST call. The function is set to status/update.xml, the HTTP method set to POST (the default is, logically, GET), and a status message is set as a parameter. We make a synchronous call, and we're done. The bonus of using OAuth to authorise with Twitter is that you get the nice "from whatever" annotations on the tweets, to promote your application.

The full source of this example is available in git, along with other examples for Flickr and Fire Eagle. If you want to understand the differences between OAuth 1.0 and 1.0a but don't fancy reading both specifications in full, I can heartily endorse An Idiots Guide To OAuth 1.0a.

NP: Simple Things, Zero 7

FAIL

if (g_list_length(nb_pages) != 0) {

FAIL

for( i=0; i < g_list_length( GTK_CLIST(clist)->selection; i++ ){
  gint row = (gint)g_list_nth_data( GTK_CLIST(clist)->selection, i);

TURBOFAIL

• Don't crash if you edit a task and then delete it
• Lots of translations
• Don't use SexyIconEntry
• Move task ellipsising to the middle
• Show tooltips for tasks with notes

As usual, download from the Pimlico Project.

Now the Eee Keyboard also has a small touchscreen by the side of the keyboard, which had generally been shown displaing a calendar and the time. Fairly useful but nothing that interesting. However, they have recently demonstrated Moblin 2 running on the Eee, including the Myzone social desktop update thingy.

Now this is pretty neat. I don't know how the touchscreen is related to the main display, but a custom Moblin 2 panel and Myzone tailored to fill the touchscreen would be really cool. Now, where can I get an Eee Keyboard from...

NP: Arecibo Message, Boxcutter

Okay, I admit at times it looks like a parody, but honestly it isn't!

• Read the track artist instead of album artist in Musicbrain3
• Don't crash if the release date is unknown
• Read tracks when falling back to gvfs

Finally, a call for someone with deep LAME knowledge. The GStreamer LAME element is, well, lame because it sets a number of properties to default values that make it very difficult for LAME to work well. Someone who understands how all of the LAME settings operate needs to sit down, vet the settings and remove the pointless ones, unset most of the rest, leaving the 'preset' setting as the only one which has a default value. At the moment there are many contradictory default settings which mean LAME produces rather badly encoded files. Any takers?

• Add --edit-task
• Use gtk_show_uri if available
• Lots of translation updates
• Add magic patterns "in x days" and "in x weeks"

As usual, download from the Pimlico Project.

• Put the disc number in the file name
• Support multiple genres
• Use libcanberra for event sounds
• Handle custom patterns in the prefs dialog
• Remove Musicbrainz data if the track data is changed
• Fix disc number editing logic
• And lots of bug fixes by many people

• Port to Brasero (Luis Medinas)
• Fix Solaris builds (Brian Cameron)
• Drop libgnome (Iain Holmes, Emilio Pozuelo Monfort)
• Fix conflicting mnemonics in the message area (Bastien Nocera)
• Fix mb3 backend (Bastien Nocera)

NetworkManager: <info>  (ttyACM0): powering up... 
NetworkManager: <info>  Registered on Home network 
an 15 10:50:04 blackadder NetworkManager: <info>  Associated with network: +COPS: 0,2,"23415" 
  NetworkManager: <WARN>  dial_done(): Dialing timed out </WARN>

Or:

NetworkManager: <info>  Activation (ttyACM0) Stage 1 of 5 (Device Prepare) complete. 
NetworkManager: <info>  (ttyACM0): powering up... 
NetworkManager: <info>  Registered on Home network 
NetworkManager: <info>  Associated with network: +COPS: 0,2,"23415" 
NetworkManager: <info>  Connected, Woo! 
NetworkManager: <info>  Activation (ttyACM0) Stage 2 of 5 (Device Configure) scheduled..
. 
NetworkManager: <info>  Activation (ttyACM0) Stage 2 of 5 (Device Configure) starting...
 
NetworkManager: <info>  (ttyACM0): device state change: 4 -> 5 
NetworkManager: <info>  Starting pppd connection 
NetworkManager: <debug> [1232015456.962700] nm_ppp_manager_start(): Command line: /usr/s
bin/pppd nodetach lock nodefaultroute user web ttyACM0 noipdefault usepeerdns lcp-echo-failure 0 lcp-echo-interval 
0 ipparam /org/freedesktop/NetworkManager/PPP/4 plugin /usr/lib/pppd/2.4.4/nm-pppd-plugin.so 
NetworkManager: <debug> [1232015456.964964] nm_ppp_manager_start(): ppp started with pid
 29590 
NetworkManager: <info>  Activation (ttyACM0) Stage 2 of 5 (Device Configure) complete. 
pppd[29590]: Plugin /usr/lib/pppd/2.4.4/nm-pppd-plugin.so loaded.
pppd[29590]: pppd 2.4.4 started by root, uid 0
NetworkManager: <WARN>  pppd_timed_out(): Looks like pppd didn't initialize our dbus mod
ule

Anyone know what the problem could be?

NP: Rendez-Vous (Mexico), Erik Truffaz featuring Murcof