Tell me more

Here is the current w3 draft:

http://www.w3.org/TR/speech-synthesis/

 Demo:

go to this page

http://www2.research.att.com/~ttsweb/tts/demo.php

and copy/paste this element:

<phoneme alphabet=”ipa” ph=”kæt”>dog</phoneme>

The body of the phoneme element can be left empty, but in this case I wrote “dog” just for fun.

Can’t we just make computers better at speaking English?

Sure, and it would be trivial to make a build jQuery plugin that converts  ALT attributes to IPA ones. But what about regional dialects and made up words like Flickr or Imgur. What about ambiguous acronyms? Should the company AAA be pronounced “ay ay ay” or “Triple A”? Is it “My S-Q-L” or “My sequel”?

Sounds good, when’s it coming out?

Fair enough, maybe you sacrifice a goat we could get it in IE11. Honestly, I don’t know. Maybe never. The spec hasn’t been touched since 2004 but I hope people start taking it seriously.

Github isn’t free, buts its very inexpensive and amazing. Inline comments, automatic issue tracking, and you can backup your code to their servers without needing commit access to the main repo.

The work flow

1) You create a repo.

2) Person A forks the repo and clones it to their local machine.

3) They work all day long and at the end of the day push their changes to their fork.

4) They feel inspired when they get home and want to keep working. They clone their fork from github to their home computer and keep working. With SVN, they would have needed to create a branch (which means they have commit access to the main repo) or they need to create their own repo which makes merging a pain later.

5) They’re done working at home, they push their changes to their fork, and submit a pull request to you.

6) You open the pull request and review their work. You notice they made a mistake in a regular expression on line 57 so you click the line and add a comment. An issue is automatically created and assigned to them.

7) They log in, see the issue, fix the issue, and resubmit the pull request.

8) You review and approve and life goes on.

A few comments

A user can submit a pull request to any fork, not just the master one. This makes forming teams super easy. You create a repo and all the members of Team A fork it, but they submit their pull requests to the Team Leader of Team A. Maybe they send daily pull requests to the Team Leader to review and he sends you weekly pull requests. You get the idea. This is perfect for our organization. Someone with a year of experience is perfectly capably of reviewing/mentoring a new hire, but you still want them to be reviewed by the people with 5 years of experience.

Github charges by the number of private repos, and forks don’t count! And yes, the fork of private repo stays private. If you have a single project using a single repo, it doesn’t matter if you have 1000 people collaborating on this project, it still only counts as one repo. Even though “technically” each fork is a repository (and stays private!) it doesn’t count towards your total.

I still think SVN is great. I have private SVN hosting with Beanstalk and I don’t plan on switching to github, but this is for my private projects. I don’t need to worry about collaboration, code review, team hierarchies, ect. The simplicity of SVN makes it an easy choice for personal projects, but for anything that requires collaboration, I’m going with github from now on.

Demo | Source Code

A little about IPv6

An IPv6 address is 128-bits and commonly represented as eight colon separated groups with each group consisting of four hexadecimal digits. For the general unicast address format, the first 64-bits are used for routing and the second 64-bits are the interface identifier (RFC 4291). Of the first 64-bits, the first 48 are used for regional routing and the next 16 are used for subnetting. Thus for practical subnet planning, you can assign blocks from /49 to /64. My calculator only lets you select a prefix up to /63 because if you can’t subnet a /64 block, you can only assign it.

How to use this calculator

Step 1)

You can enter the network IP to be subnetted as a full address or in the shortened form with leading zeros removed and consecutive groups of zeros removed.

Either 2001:09fe:000a:0000:0000:0000:0000:0000 or 2001:9fe:a:: is fine.

Step 2)

Enter the number of subnets you want created at each level and press “Add Level.” Example: your university has 10 colleges, each college has 3 buildings, and each building has 8 departments. Add 10, then 3, then 8.

Step 3)

Build it! Then use the + button to show / hide subnets. The position string on the left hand side should help you identify who the network is subnetted to (eg. the subnet in position 4.2.3 is the 3rd department in the second building of the 4th college).

Step 4)

Post questions in the comments and bugs on GitHub.

xdg-mime default fedora-deluge.desktop x-scheme-handler/magnet

Why isn’t Compiz in the Fedora 17 repos?

Compiz is a replacement window manager for the GNOME window manager called Metacity. When GNOME 3 came out last year, they replaced Metacity with Mutter. Apparently there are library incompatibilities that prevent Compiz from replacing Mutter. I didn’t look into the details because we don’t care about GNOME issues, we’re XFCE users!

I’ll also note that Red Hat is currently reviewing the situation and might bring Compiz back. Who knows. I for one think it’s silly to drop a package just because it doesn’t work with one desktop environment out of many.

Why I love Compiz (you can skip this)

I like Compiz a lot and not because of the silly effects. The #1 feature to me is the grid plugin. I press Ctrl+Num7 and my window auto resizes to the top left corner. I press Ctrl+Num4, it resizes to the whole left half of the screen. If I press Ctrl+Num4 twice it takes up 1/3 of the left hand side and then I press Ctrl+6 3 times on another window and that one takes up 2/3 of the right half of the screen. You get the idea. Compiz is the best tiling window manager I’ve every used and it’s not even a tiling window manager.

It’s awesome, it makes you way more productive, and nothing even comes close to rivaling it. I could live without every other part of compiz (though the slick effects are nice) but the grid makes it worth it.

Installing Compiz

First you need to download the relevant rpm packages from the Fedora 16 repository. I’d suggest making a new directory called “compiz-rpms” or whatever you want.

Required Packages

Shout out to Brombor on the Fedora forums for making this list. All the linked packages below are for the x86_64 version. (Link to the i386 repo if you need it).

• boost-serialization
• libcompizconfig
• ccsm
• compiz
• compiz-bcop
• compiz-fusion-extras
• compiz-fusion-extras-gconf
• compiz-gconf
• compiz-gnome
• compiz-gtk
• compiz-manager
• compiz-plugins-main
• compiz-plugins-main-gconf
• compizconfig-backend-gconf
• compizconfig-python

Or you can be a cool kid and just do

wget -v http://archives.fedoraproject.org/pub/fedora/linux/releases/16/Everything/x86_64/os/Packages/boost-serialization-1.47.0-3.fc16.x86_64.rpm http://archives.fedoraproject.org/pub/fedora/linux/releases/16/Everything/x86_64/os/Packages/libcompizconfig-0.9.5.0-2.fc16.x86_64.rpm http://archives.fedoraproject.org/pub/fedora/linux/releases/16/Everything/x86_64/os/Packages/ccsm-0.9.5.0-1.fc16.noarch.rpm http://archives.fedoraproject.org/pub/fedora/linux/releases/16/Everything/x86_64/os/Packages/compiz-0.9.5.0-4.fc16.x86_64.rpm http://archives.fedoraproject.org/pub/fedora/linux/releases/16/Everything/x86_64/os/Packages/compiz-bcop-0.8.8-1.fc16.noarch.rpm http://archives.fedoraproject.org/pub/fedora/linux/releases/16/Everything/x86_64/os/Packages/compiz-fusion-extras-0.9.5.0-3.fc16.x86_64.rpm http://archives.fedoraproject.org/pub/fedora/linux/releases/16/Everything/x86_64/os/Packages/compiz-fusion-extras-gconf-0.9.5.0-3.fc16.x86_64.rpm http://archives.fedoraproject.org/pub/fedora/linux/releases/16/Everything/x86_64/os/Packages/compiz-gconf-0.9.5.0-4.fc16.x86_64.rpm http://archives.fedoraproject.org/pub/fedora/linux/releases/16/Everything/x86_64/os/Packages/compiz-gnome-0.9.5.0-4.fc16.x86_64.rpm http://archives.fedoraproject.org/pub/fedora/linux/releases/16/Everything/x86_64/os/Packages/compiz-gtk-0.9.5.0-4.fc16.x86_64.rpm http://archives.fedoraproject.org/pub/fedora/linux/releases/16/Everything/x86_64/os/Packages/compiz-manager-0.6.0-13.fc15.noarch.rpm http://archives.fedoraproject.org/pub/fedora/linux/releases/16/Everything/x86_64/os/Packages/compiz-plugins-main-0.9.5.0-2.fc16.x86_64.rpm http://archives.fedoraproject.org/pub/fedora/linux/releases/16/Everything/x86_64/os/Packages/compiz-plugins-main-gconf-0.9.5.0-2.fc16.x86_64.rpm http://archives.fedoraproject.org/pub/fedora/linux/releases/16/Everything/x86_64/os/Packages/compizconfig-backend-gconf-0.9.5.0-1.fc16.x86_64.rpm http://archives.fedoraproject.org/pub/fedora/linux/releases/16/Everything/x86_64/os/Packages/compizconfig-python-0.9.5.0-1.fc16.x86_64.rpm

Install boost-serialization

Once you install boost-serialization, you have to prevent it from updating. Install the version lock plugin and boost-serialization.

yum install yum-plugin-versionlock
yum localinstall boost-serialization-1.47.0-3.fc16.x86_64.rpm

Lock boost-serialization

yum versionlock boost-serialization

Install everything else

yum localinstall libcompizconfig
yum localinstall ccsm
yum localinstall compiz*

You could technically run compiz at this point and have wobbly windows and all that stuff, but you’d be missing one important thing… a window decorator. All your window buttons (close, minimize, ect) will be gone. So unless you feel like holding Alt to move windows and using shortcuts to close them, I’d suggest installing emerald.

Installing emerald

You can’t just install emerald from the Fedora 16 repo because it conflicts with compiz-plugins-main.

Option 1: Use my RPM

I already built the emerald package for my own use so I’ve uploaded the rpm to my github (download here). If you’re running the XFCE spin of Fedora 17 x86_64, then you should be able to just download this and install it with

rpm -ivp emerald-0.9.5-0.1.rpm

If this works, then you’re in business and can skip to the section “Running Compiz.” But remember that if you have issues with the window bar not showing up then you should remove my emerald package and build your own.

Option 2: Compile emerald from SRPM file

An SRPM file is the source code + SPEC file (build instructions). It usually has the extension .src.rpm. I’ve uploaded the SRPM of emerald-0.9.5-0.1 to my github account. SRPM download

How to create an RPM package

Here are the instructions that I followed from the Fedora wiki
http://fedoraproject.org/wiki/How_to_create_an_RPM_package

I’ve provided abbreviated instructions below, but if you have problems then you should consult the wiki guide.

Step 1: Compiz-Devel
To build emerald you need to install compiz-devel and to install compiz-devel you need install some other packages. So lets get that out of the way.

yum install libwnck-devel gtk2-devel intltool libtool gettext-devel libXres-devel

Download and install compiz-devel

wget http://archives.fedoraproject.org/pub/fedora/linux/releases/16/Everything/x86_64/os/Packages/compiz-devel-0.9.5.0-4.fc16.x86_64.rpm
yum localinstall compiz-devel-0.9.5.0-4.fc16.x86_64.rpm

Step 2: Preparing your system
Install some core development tools

yum install @development-tools
yum install fedora-packager

You never want to create packages as root. We’re going to create a new user named makerpm, add the user to the ‘mock’ group, set a password, and login as that user

/usr/sbin/useradd makerpm
usermod -a -G mock makerpm
passwd makerpm
su - makerpm

Make sure you’re logged in as the dummy user and in their home directory. Then create the required directory structure.

rpmdev-setuptree

Step 3: SRPM to RPM
Chances are that the emerald SRPM you downloaded is owned by your user account. Move it to the homedir of your dummy account and chown it to them.

Create a directory and unpack the SRPM

mkdir emerald_src_rpm
cd emerald_src_rpm
rpm2cpio ../emerald-*.src.rpm | cpio -i

Let’s just verify the spec file real quick

rpmlint emerald.spec

Create binary RPMS

rpmbuild -ba emerald.spec

If successful, the RPMS will be created within ~/rpmbuild/RPMS/

Lets verify that the build worked correctly

rpmlint emerald.spec  ~/rpmbuild/RPMS/*/emerald*.rpm

I got a few warning but no errors. If everything went well for you, enter the ~/rpmbuild/RPMS directory and into the architecture subdirectory. Install your new emerald package from root with

rpm -ivp emerald*.rpm

Running Compiz

Enable plugins

You need to activate a few plugins that provide basic window manager behavior or else you will have no ability to drag, scale or close any windows as soon as compiz is activated. These plugins should be enabled by default, but lets be sure. Open the Compiz Settings Manager

ccsm

and put check marks next to “Window Decoration” under Effects and “Move Window” & “Resize Window” under Window Management.

Replace xfwm4 with Compiz

Easy peasy. To start Compiz and Emerald together, just run

compiz-manager &

What if something goes wrong?

Window decorator isn’t appearing
Switch back to xfwm4.

xfwm4 --replace &

Screen locked up
Ctrl+Alt+Backspace will automatically log out you.

Taskbar and Desktop not appearing
The screen locked up, you ctrl+alt+backspaced out, and now every time you log in you get the same broken Compiz session. You need to clear your session cache.

rm ~/.cache/sessions/*

Start Compiz Automatically

After you’ve verified that Compiz/Emerald is working correctly, you can set it to start automatically when you log in.

Start > Settings > Session and Startup > Application Autostart

Add a new item
(Name) Compiz
(Description) Start Compiz
(Command) compiz-manager

Now Compiz should start automatically, every time you log in.

Note: If something breaks and you need to stop Compiz from starting automatically but you can’t access the GUI, this is how you disable application autostarts from the command line:

cd ~/.config/autostart/
rm Compiz.desktop

Bug fixes

I’m just going to post random bug fixes as I encounter them (hey, I never told you this was stable).

Tooltips and Menus not showing correctly

There’s a weird bug with XFCE and Compiz where tooltips don’t display correctly some times. Open up the Compiz Settings Manager (ccsm), scroll down to “Effects”, and enable “Animations.” Now tooltips display correctly every time!

Compiz crashes with the error “/usr/bin/compiz-manager: line 367: Segmentation fault”

It was to do with the grid plugin. Open ~/.config/compiz-1/compizconfig/Default.ini and remove “[grid]” and everything under it.

Compiz-manager won’t start after installing the latest video drivers, “No whitelisted driver found”

I installed the latest ATI Radeon video drivers and now compiz-manager refuses to start because my drivers aren’t in the whitelist. Rather than use compiz-manager to start compiz and emerald, we can start the both manually.

LIBGL_ALWAYS_INDIRECT=true compiz --replace --ignore-desktop-hints ccp & emerald --replace &

Setting LIBGL_ALWAYS_INDIRECT to true is necessary if you’re running AIGLX.
I’m not sure what –ignore-desktop-hints does, but its a recommend option on compiz.org
ccp uses compizconfig for settings (so you have window decorators and stuff)

Make Compiz start automatically again
Try out the above command and verify that it works.
Remove your XFCE Autostart entry via the GUI or deleting the .desktop file in ~/.config/autostart/
Add the above command to your ~/.bash_profile file
Log out and in again

This is the gold standard of iPad web frameworks. It’s amazing (and free!), check out the kitchen sink demo.

Project Page

Slablet

Project Page

Demos

WebApp

Project Page

Demo

Splitview

http://asyraf9.github.com/jquery-mobile/

jQuery Mobile

http://jquerymobile.com/

https://play.google.com/store/apps/details?id=com.colinrrobinson.draughts

I’ve been hovering around 95%-100% for a few weeks now. Being at 95% is fine, but 100% can cause all kinds of problems. I was testing a form on my local web server and couldn’t do file uploads because “the disk was full.” I was writing code and my editor couldn’t save the file because “the disk was full.” Its frustrating because you have hundreds of free gigs where you’re saving the file, but because rootfs is full, you can’t!

I’ll start by showing you the output of df

$ df
Filesystem     1K-blocks     Used Available Use% Mounted on
rootfs          10209220  9625988     71164 100% /
/dev             4055116        0   4055116   0% /dev
run              4058036      252   4057784   1% /run
/dev/sda3       10209220  9625988     71164 100% /
shm              4058036      796   4057240   1% /dev/shm
tmpfs            4058036     3160   4054876   1% /tmp
/dev/sda1          99590    34166     60205  37% /boot
/dev/sda4      449789792 23726068 403549860   6% /home

Not good, but its my own fault. I only allocated about 10g to /dev/sda3 when I setup the OS. I should I have predicted that I would need more.

The real solutions is to boot to a livecd and resize /dev/sda3. This can take a long time, break your grub install, and cause other headache inducing effects.

Here are some tricks I’ve picked up to get your rootfs down from the dreaded 100% to something you can actually work with.

Clear your pacman cache

When pacman downloads a package for you, it keeps a version in its cache (usually /var/cache/pacman/pkg). This is great if you ever want to downgrade a package, but its not necessary and it takes up space.

Clearing your cache is as easy as:

$ pacman -Sc

There are more complicated scripts on the net which do things like manage how mange revisions back to save packages for, but this is fine for now. Lets take a look at the difference that made

$ sudo df
Filesystem     1K-blocks     Used Available Use% Mounted on
rootfs          10209220  9303916    393236  96% /

Down to 96% already! This is all the difference in the world, now I can actually save this document I’m writing. Lets see how much further we can get this down.

Remove software you don’t use

Now might be a good time to remove unused software. I normally run a pretty tight system anyways, but I still found room to improve.

For example, I use Xfce as my Desktop Environment but I use Compiz as my Window Manager rather than xfwm4. This let me remove xfwm4 and xfwm4-themes and shed a few MBs. I originally installed both VirtualBox and VMware but I only use the latter. See you later Vbox!

Kill the orphans!

Or rather… remove orphaned packages. First do a pacman query -Q for all dependencies -d that are unrequired -t.

$ pacman -Qdt

If you have unrequired dependencies, why not remove them. This simple command with recursively remove all unrequired dependencies.

pacman -Rs $(pacman -Qtdq)

Where are we at?

$ df
Filesystem     1K-blocks     Used Available Use% Mounted on
rootfs          10209220  9090832    606320  94% /

BleachBit

Similar to CCleaner on Windows, BleachBit clears out temp files, system caches, and other junk. You can find it here in the AUR (There is also a CLI version).

Figure out what’s taking up space

Still not satisfied with the size of rootfs? Get down into the file structure and find out what’s taking up space. The arch wiki has a list of disk usage display programs. I’m personally a fan of ncdu (home page) (package details).

Have a suggestion of your own? Still having problems with your file system? Make a comment!

Most code review software is designed for pre-commit comments. A reviewer looks at the code before it gets added to the version control system and then approves or denies the changes. This is obviously vital for large projects involving multiple developers, but what about small shops where people work mostly independently and on different projects? What if you don’t have the resources or the need for an extensive review process and would rather casually browse your colleges code to add constructive criticisms or notify them of best practices which they might be unaware of?

To me the ideal tool will interface with subversion and allow reversion specific and line specific post-commit comments. I want to be able to browse our source code whenever I feel like it, write a comment about a specific line, and then have the author notified of the comment.

The PeerReviewPlugin for Trac looked like a winner, until I read this review

PeerReview Screenshot

In my company, we looked briefly at the “peerreview” plugin on TracHacks, and were very disappointed with it.

It seems obvious to us that a code-review plugin would naturally default to assuming that an entire Subversion commit should be code-reviewed. Unfortunately, the peerreview plugin forces you to manually identify the lines of code that you want to review. It doesn’t even give you hints in which lines might have changed with a particular commit, which means that if you don’t enter the line that changed carefully, you could end up re-reviewing the same lines of code over and over.

It still looks promising, and I don’t expect to find an out-of-the-box solution that is perfect for my specific situation. Right now ReviewBoard is looking up and I could always modify the PeerReview plugin to suit my needs (hooray, opensource). Please, leave a comment about how you and your company do code review. What code review systems have you worked with? What other solutions might I have passed over?