Hi. I'm Paul Rouget. I work for Mozilla. I have fun with the web and the Mozilla platform. I write code. I give talks. I tweet.
This is my blog.
Hi. I'm Paul Rouget. I work for Mozilla. I have fun with the web and the Mozilla platform. I write code. I give talks. I tweet.
This is my blog.
Tuesday, 07 September 2010
Ok, comments are here (powered by Disqus). I think my blog engine is ready now :)
Thursday, 26 August 2010
Monday, 23 August 2010
Create a directory, let's say "myvideos". All the comming commands will make sure that everything stays in this directory. Your system won't be altered.
dependencies:
git clone git://review.webmproject.org/libvpx.git
mkdir ~/myvideos/dist
./configure --prefix=~/myvideos/dist && make && make install
Let's build ffmpeg:
$ LDFLAGS=-L$~/myvideos/dist/lib \ CFLAGS=-I~/myvideos/dist/include \ ./configure --prefix=~/myvideos/dist --enable-gpl\ --enable-nonfree --enable-libvpx --enable-libvorbis\ --enable-pthreads --enable-libx264 --enable-libfaac\ --enable-libtheora $ make && make install
~/myvideos/dist/bin/ffmeg
But first, you need to tell where to find libraries:
export LD_LIBRARY_PATH=~/myvideos/dist/lib/
Here are my 3 command lines to convert to WebM, OGG/Theora and mp4:
ffmpeg -i input.mov \
-acodec libvorbis -ac 2 -ab 96k -ar 44100 \
-b 345k -s 640x360 output.ogv
ffmpeg -i input.mov \
-acodec libvorbis -ac 2 -ab 96k -ar 44100 \
-b 345k -s 640x360 output.webm
ffmpeg -i input.mov \
-acodec libfaac -ab 96k \
-vcodec libx264 -vpre slower -vpre main \
-level 21 -refs 2 -b 345k -bt 345k \
-threads 0 -s 640x360 output.mp4
Hope it helps.
Monday, 23 August 2010
html5test is a good test. It's easy to understand. Neils found a good way to rate HTML5 support. It's not perfect, but I think a perfect test is not possible if you want to keep it simple. But there's still room for improvements.
Look at the current version of html5test, the Forms part.
Webkit has a good score (34/38, 89%), although their HTML5 forms support is incomplete (no UI, validity barely implemented, but they expose the input type). The way forms are tested is not that accurate:
just testing if the input type is supported as a DOM property is not enough.
The good way is to check that the constraint is actually applied (input.validity.valid).
//Exemple: testing <input type=email>
/*
* How it was tested
*/
var valid = (input.type == "email");
/*
* A better way to test it
*/
input.value = "foo";
var valid = false;
if (!input.validity.valid) {
input.value = "foo@bar.org";
if (input.validity.valid) {
valid = true;
}
}
I also added some tests for input.form, input.labels and label.control. I'm also testing the forms related CSS selectors (:required, :optional, :valid and :invalid).
I sent a patch to Neils. You can see the result here beta.html5test.com. Basically, webkit score is less good now for the forms section (29/45, 64%), but still better than Firefox4 ;) (and still very good for Opera).
Firefox4 nightlies now support part of the HTML5 form specification (talk to Mounir, the main developer). The latest nightly's is 24/45 (53%), and it'going to get better :)Thursday, 19 August 2010
The Nanonote, a very, very small laptop, running Linux. Just bought it from Hackable-device. Can you see it? :)
Thursday, 19 August 2010
Now, the RSS feed should include the content of blog posts.
Step after step, I'm building my blog engine :)
Thursday, 05 August 2010
Important edit: From the different results you shared with me: Sounds like they are probably using a D2D enabled build. Also, the result of this test is not constant at all (I just had a "111" on IE9). The only thing this test teachs us is that a hardware accelerated browser is faster. I will publish more data soon.
Honestly, I'm super excited to see IE9 coming with so many cool features. It's a little revolution :)
Microsoft, you do it right (well, I would like to see IE9 for Windows XP as well).
I just want to add one note to their awesome performance tests. When they test Firefox
they don't activate the D2D acceleration in Firefox (and I understand that, we don't enable it by default, yet).
But then, the result is a bit different...
Friday, 30 July 2010
For the last few weeks, I've been using the same HTML template as slides for my presentations. This template is now online. It works great with Firefox 4 and I'm trying to make it work with other browsers (need help here). Feel free to grab the code and play with it:
Thursday, 29 July 2010
Ok, so now, the RSS should work (paulrouget.com/index.xml). A part of this RSS is generated with gcc :)
gcc -traditional -E -P -C -x c -D VAR=value
Thursday, 29 July 2010
I'm working on it...
Thursday, 29 July 2010
You know, I'm a huge fan of Lenovo products. Like "HUUUGE". I love the Thinkpads.
But, seriously, guys, the Lenovo Shop is just the most awful online store ever. They don't know HTML links (just JS jumps), you get lost in like 2 minutes, it's a mess to understand the difference between all the products... well, a nightmare.
What about a sober but efficient design, like you do for your products?
Thursday, 29 July 2010
I use to have a Gandi blog (dotclear powered). Gandi did an awesome job, but they don't allow me to use my own HTML (I want the <video> tag!). So I decided to create my own blog. It's based on a shell script and some Vim magic tricks. No RSS for now, and no comments (for ever, don't want to deal with spam).