Stuff to remember.

moved to gmarik.info

2009-04-02T04:17:00.001+02:00

This blog will stay but won't be updated.

I've moved to gmarik.info

Update your bookmarks! :)

Code snippets - gists

2009-01-02T02:50:00.003+02:00

If you collaborate and share code snippets you may like gist.sh

Description:


Bash script for getting/posting code snippets (gists) to/from http://gist.github.com

Usage


    * Getting gist.sh:

      $ curl http://github.com/gmarik/gist.sh/raw/master/gist.sh > gist.sh && chmod 755 gist.sh && sudo mv gist.sh /usr/local/bin/

    * Posting to GitHub:

      $ cat file | gist.sh

or

$ gist.sh < file·

    * Getting from GitHub:

      $ gist.sh 1234

Moved to log.gmarik.info

2008-12-22T02:32:00.009+02:00

Since now this blog is accessible at log.gmarik.info !

Log:

registered a domain with google apps
Disabled both Sites and Start Page services using google apps Service settings (to avoid nasty Another blog is already hosted at this address error)
used www.gmarik.info as domain name for custom domain name (advanced mode in blogger.com blog's Settings/Publishing )

Nice!

PS: access google apps domain managing dashboard at www.google.com/a/your_domain_name

Linux howto: Enable Java runtime in Firefox

2008-11-25T13:10:00.007+02:00

You need to
1) have jre installed
2) Edit > Preferences / [ Content ] tab: [ ] Enable Java checkbox checked
3)
$ cd /path/to/firefox/plugins
$ ln -s /path/to/java/jre/plugin/i386/ns7/libjavaplugin_oji.so ./
4) test

source

PS:
running .jnlp apps:
$ $JAVA_HOME/bin/javaws ./app.jnlp

Ubuntu: sysv-rc-conf - missing admin tool

2008-11-10T13:49:00.004+02:00

I'm coming from RedHat school.
After switching to Ubuntu I missed chkconfig tool for managing services.

But Ubuntu provides same managing interface with sysv-rc-conf instead.

$sudo apt-get install sysv-rc-conf

Nice!

Peoplenet on Ubuntu Linux

2008-10-29T01:12:00.005+02:00

I've got CCU-550(3G/UMTS) modem.
I've never installed modems on Linux before so I just hoped that newest distros have right tools to do that.

I wanted to start using it quickly therefore began with NetworkManager's PointToPoint connection option. But it didn't want to work...

Then I looked at pppd stcripts. But it didn't work for me either...

At that point I didn't have internet access so the only option i could progress with was reading man pages for wvdial.

Interesting what man page says about wvdial - PPP dialer with built-in intelligence.
And it turned out to be true :)

So i launched


$man 5 wvdial.conf

copied sample configuration to /etc/wvdial.conf and modified to look like:


[Dialer Defaults]
Modem = /dev/ttyACM0
Baud = 460800
Init = ATZ
Phone = #777
Username = 80921111111@people.net.ua
Password = 111111

To get the proper Modem location i did:


$dmesg |grep tty
[  164.384650] cdc_acm 2-1:1.0: ttyACM0: USB ACM device

Then launching:


$wvdial

I got it working:


$ ifconfig |\grep -A 1 ppp 
ppp0      Link encap:Point-to-Point Protocol  
          inet addr:my.ip.addr.ess  P-t-P:pp.ip.addr.ess  Mask:255.255.255.255

Wow! Nice! :)

using rake with sudo

2008-10-25T17:22:00.012+02:00

I was installing datamapper gems like:


gmarik@gm:master\dm-more.git$ sudo rake install
sudo: rake: command not found

sudo restricts environment settins:


$ sudo env|\grep PATH
PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin

As you can see rake path is not included:


$ which rake
/usr/local/lib/ruby/gems/bin/rake

You can edit sudo PATH to include /usr/local/lib/ruby/gems/bin.
=== UPDATED AT 2008.12.04 =====
or run rake with PATH set like:


sudo env PATH=$PATH rake

===============================
Bu i just symlinked rake:


$sudo ln -s `which rake` /usr/local/bin/

Now it works:


$ sudo rake --version
rake, version 0.8.3

Ubuntu 8.04: fglrx-ATI direct rendering, issue 3: Kernel update

2008-10-20T11:00:00.007+02:00

Just rebooted my machine after kernel update few days ago.
Previous Kernel: 2.6.24-19-generic
Current version:


$ uname -a
Linux gm 2.6.24-21-generic #1 SMP Mon Aug 25 17:32:09 UTC 2008 i686 GNU/Linux

And, yes - got fglrx broken again:


$ fglrxinfo 
display: :0.0  screen: 0
OpenGL vendor string: Mesa project: www.mesa3d.org
OpenGL renderer string: Mesa GLX Indirect
OpenGL version string: 1.4 (2.1 Mesa 7.0.3-rc2)

$ glxinfo |grep direct
3:direct rendering: No (If you want to find out why, try setting LIBGL_DEBUG=verbose)

As for solution - check out my prevous post for same issue

enjoy!

Ubuntu 8.04: ATI directr rendering, issue 2

2008-06-17T19:42:00.006+02:00

Well i got it broken again.

This time running
$ sudo modprobe --ignore-install fglrx
yielded
FATAL: Module fglrx not found

WTF??

By examining System > Administration > Hardware Drivers i found no driver for ATI card.

To fix that i went to System > Administration > Synaptic Package Manager and searched for fglrx.

I had linux-restricted-modules installed for 2.6.24.12-16.34 kernel installed.
But actual kernel version was

$ uname -a
Linux gm 2.6.24-19-generic #1 SMP Wed Jun 4 16:35:01 UTC 2008 i686 GNU/Linux

So removing old and installing appropriate linux-restricted-modules for 2.6.24.13-19.24 kernels solved my direct rendering issue:

$ glxinfo |grep direct
3:direct rendering: Yes

PS. Don't forget to reboot!
Hope this will help you too! :)

Ubuntu 8.04: ATI direct rendering issue

2008-05-05T13:24:00.004+02:00

After upgrade to Ubuntu 8.04 i got direct rendering broken:


$ glxinfo |grep direct
3:direct rendering: No

Though everything was configured properly and seemed to work ok.

Removing MESA drivers solved the issue.

Overall: 8.04 is still kinda slow comparing to 7.10. (Sometimes I think I shouldn't have upgraded to 8.04)

DataMapper: installing do_mysql on Ubuntu 8.04

2008-04-30T11:12:00.010+02:00

Usually I had gems installed flawlessly.

But this time i wasn't so lucky with do_mysql.

I got this:


$ sudo gem install do_mysql --no-ri --no-rdoc
Building native extensions.  This could take a while...
ERROR:  Error installing do_mysql:
 ERROR: Failed to build gem native extension.

/usr/local/bin/ruby extconf.rb install do_mysql --no-ri --no-rdoc
checking for mysql_query() in -lmysqlclient... no
checking for main() in -lm... no
checking for mysql_query() in -lmysqlclient... no
checking for main() in -lz... no
checking for mysql_query() in -lmysqlclient... no
checking for main() in -lsocket... no
checking for mysql_query() in -lmysqlclient... no
checking for main() in -lnsl... no
checking for mysql_query() in -lmysqlclient... no
*** extconf.rb failed ***
Could not create Makefile due to some reason, probably lack of
necessary libraries and/or headers.  Check the mkmf.log file for more
details.  You may need configuration options.

Little extra configuration solved the issue


$ sudo gem install do_mysql --no-ri --no-rdoc -- --with-mysql-lib=/usr/lib/mysql/

Now i can play with DataMapper eventually :)

Ruby's here document mini tutorial.

2007-12-27T23:17:00.000+02:00

Last update: 2007.12.30

Basics

Here Document(or HereDoc) is a way to declare String literal in Ruby programming language:


some_text = <<END_OF_STRING
 You can write any text here for your document that's why such
 statement is called - HereDoc
END_OF_STRING

That's it! Now some_text variable points to the string object containing the text going between END_OF_STRING
As you may know HereDoc isn't a unique Ruby feature, rather it's a common construct(with some distinctions) for scripting languages "brewed" in Unix environment.

The terminator

By Ruby convention a variable starting with capital letter is a constant. But that's not a case for the END_OF_STRING from previous example, because terminator is just a string which parser treats as the end of HereDoc.
Well if a terminator is a string then can i use arbitrary(say put spaces within) string as the terminator like end of string? The answer is - yes you can!
<<heredoc is interpreted same as <<"heredoc" (please note double quotes around latter heredoc).
But explicit notation(with quotes) is a bit more powerful.

String interpolation

By explicitly enclosing terminator with quotes you may have:

a_text = <<"Ruby, please end my HereDoc once you find this terminator string"
Hello world!
Ruby, please end my HereDoc once you find this terminator string

fuzzy_names = <<">>"
foo, baz, bar
>>

Wow, if i can use double quoted string, then probably i can use single quoted string as well:

puts <<'end of string'
 1+1=#{1+1}
end of string

prints

1+1=#{1+1}

as single quoted strings aren't interpolated unlike double quoted:

puts <<"end of string"
 1+1=#{1+1}
end of string

prints

1+1=2

Indent modifier

By default HereDoc terminator is expected to be placed on the very beginning of the separate line
By using - on HereDoc declaration, you may indent end terminator arbitrary:

greeting = <<-"here document ends"
                 Hello world
               here document ends

Keep in mind that leading spaces are kept.

Advanced usages

a, b = <<'EOA', <<-EOB
string_a
EOA
string_b
   EOB

is equal to

a, b = "string_a\n", "string_b\n"

At this point i'm thinking about HereDoc as "placeholder" that gets substituted with the string itself. Why is that important? Because you may then treat HereDoc declaration as the actual string and send messages(call methods):

<<'heredoc'.reverse == "\n!dlrow olleH"
Hello world!
heredoc

is a true statement!

Pragmatic ajax activity indicator with prototype and css

2007-12-20T20:03:00.000+02:00

Step1

Put this javascript snipped somewhere into your html or as javascript inclusion:


Ajax.Responders.register({
  onCreate: function() {
    if($$('body').first() && Ajax.activeRequestCount > 0)
        $$('body').first().addClassName('busy')
  },
  onComplete: function() {
    if($$('body.busy') && Ajax.activeRequestCount == 0)
        $$('body.busy').first().removeClassName('busy')
  }
});

Previous chunk adds a body css class name to the html's body element everytime the Ajax.Request is made to the server. Pretty simple, eh?

Step 2

Define such css styles somewhere:

 
body.busy * {
  cursor: progress !important;
}

This css rule says to change mouse cursor to convey us to wait a bit...

Step 3

Enjoy!

VirtualBox seamless mode

2007-11-15T22:55:00.000+02:00

VirtualBox's Seamless mode is just amazing feature.
You just have to check it out!

1. Boot your guest OS(windoze only i guess)
2. Devices > Install Guest Additions
3. FOllow the instructions
4. Machine > Seamless Mode

Enjoy!

VirtualBox rox!

Howto: run both 1.* and 2.* capistrano versions

2007-11-15T21:48:00.000+02:00

Assuming you already have 1.* capistrano version.
To make sure you'll need to:

$ cap --version
Capistrano v1.4.1

then query for required 2.* version:

$ gem search capi -r
*** REMOTE GEMS ***
capistrano (2.1.0, 2.0.0, 1.4.1, 1.4.0, 1.3.1, 1.3.0, 1.2.0, 1.1.0)
   Capistrano is a utility and framework for executing commands in
   parallel on multiple remote machines, via SSH.

and install:

$ gem install capistrano -v 2.1.0
Successfully installed capistrano-2.1.0

Now check capistrano version again:

$ cap --version
Capistrano v2.1.0

Note that now we have capistrano 2.* by default

Now we need to get runner for capistrano 1.*:

$cp `which cap`{,2}
creates cap2 executable

For capistrano 1.* we need to

$vim `which cap`

and change line

version = "> 0"
to
version = "< 2"

Now check version again:

$ cap --version
Capistrano v1.4.1

To have capistrano 2.* it's required to
$vim `which cap2`
and change line

version = "> 0"
into
version = ">= 2"

Now check:

$ cap2 --version
Capistrano v2.1.0

PS: ofcourse you can use "official" way for having 2 version

gruff 0.2.9 side_bar (stacked) patch

2007-11-06T23:46:00.000+02:00

I've extracted gruff ~~patch~~ (applied, thanks Geoffrey) which fixes sidebar to look like side-bar and not a stacked-bar.

VirtualBox on Fedora installation

2007-10-23T16:57:00.000+02:00

I was setting up VirtualBox on my machine and encountered error:

$ sudo /etc/init.d/vboxdrv setup
Starting VirtualBox kernel module FATAL: Error inserting vboxdrv (/lib/modules/2.6.22.9-61.fc6/misc/vboxdrv.ko): Invalid module format

Well that was because my kernel-sources didn't match my running kernel the module was built for.

I checked the /usr/src/kernels:
2.6.22.9-61.fc6-i586

and compared with uname -a:
Linux gm 2.6.22.9-61.fc6 #1 SMP Thu Sep 27 18:48:03 EDT 2007 i686 i686 i386 GNU/Linux

machine architecture uname -i:
i686

That means that to get module compiled i need to have kernel-sources(package named kernel-devel on fedora): for i686 architecture, and i had i586 instead

so find appropriate package:
$ sudo yum search kernel-devel:
kernel-devel.i686 2.6.22.9-61.fc6 updates
Matched from:
kernel-devel

then install it
$ sudo yum install kernel-devel-2.6.22.9-61.fc6.i686

now link like:
$ cd /usr/src/
$ sudo ln -s kernels/2.6.22.9-61.fc6-i686/ linux

to get linux link to point installed sources this:
[gmarik@gm src]$ ls -l
kernels
linux -> kernels/2.6.22.9-61.fc6-i686/
redhat

Having done that now we can try to setup VirtualBox again:
$ sudo /etc/init.d/vboxdrv setup
Stopping VirtualBox kernel module [ OK ]
Recompiling VirtualBox kernel module [ OK ]
Starting VirtualBox kernel module [ OK ]

PS: make sure your user have been added to vboxusers group
PPS: usually after rebooting machine i need to do:
$ sudo chmod g+rw /dev/vboxdrv

yet another html+css table bars|charts

2007-10-04T11:10:00.000+02:00

Download & view

rails: optional locals values in partials

2007-09-27T18:42:00.000+02:00

As you may know, rails' partial is a proper way for producing reusable view chunks

But sometimes you may need the partial to look different depending on contetext( say partial "_post" is used in "posts_list" and "single_post" templates). No problems - rails gives you :locals option

.rhtml


<%= render :partial => 'post', :locals => {:look_different => true} %>

And then you can do in parial like
_post.rhtml


<% if look_different %>
  <p>look different <p>
<% else %>
  <p>look usual <p>
<% end %>

But it gets painful passing all the locals all the time...otherwise you'll get NameError for not passing partial's option

Solution: optional locals values

Core rails team solution

just use local_assigns[:key] to access your local variables :)
Like
_post.rhtml


<% if local_assigns[:look_different] %>
  <p>look different <p>
<% else %>
  <p>look usual <p>
<% end %>

~~from rails 1.2.3 sources~~
from ActionView::Base docs

Google code uses jQuery!

2007-09-17T16:24:00.000+02:00

Today, surfing around google code i accidentally stoped page loading and found pretty interesting js error:


a.feed.entries[b],e=r(d),f=g(m,e);
jQuery("#techtalks .posts").append(f)
...

Meaning that Google uses jQuery!

jQuery rox! :)

The button element

2007-09-14T01:10:00.000+02:00

Here's my fix for a IE button hell(first 2 issues) i've created a while ago actually...

Useful links to read about the Button element:

Rediscovering the button element

Button the forgotten element presentation transcript

Button at Digital-Web mag.

Rails: collection index within partial render

2007-09-02T03:22:00.000+02:00

Tonight i've found a tiny and nice feature i was missing sometimes...(in order to have it created custom helper)

So when you have:
.rhtml:


 <ul>
   <%= render :partial => 'games/game', :collection => @games %>
 </ul>

you may sometimes need to access collection index. And you can easily do that by using local variable named by such convention <partial_name>_counter
Like this:

.rhtml:


 <li> <%= game_counter %> item
 </li>

PS. and yes you can always use ordered(numbered)lists(ol instead ul)

Learning Rails, issue 1

2007-01-15T21:07:00.000+02:00

I'm reading "Agile development with Ruby on Rails".
That's the coolest technical book i've been reading recently.

I decided to document issues i have during reading, just stuff to remember :)

So the very first issue I had was error during script/genration running.
The issue looks like:


Errno::ENOENT: No such file or directory - /tmp/mysql.sock

Not so much helpful...Likely to be configuration issue.
Next time I run generator with --trace option again to see the stack and to try to find out the reason.

On the very top there's a ../mysql.rb:104, lets look at it:


 if (host == nil or host == "localhost") and defined? UNIXSocket then
      unix_socket = socket || ENV["MYSQL_UNIX_PORT"] || MYSQL_UNIX_ADDR
      sock = UNIXSocket::new(unix_socket) # line 104

So it tries to use socket, then ENV["MYSQL_UNIX_PORT"], then MYSQL_UNIX_ADDR (which actually is a /tmp/mysql.sock if you look at the beginning of the mysql.rb). Here's our problem(you may want to dig deeper into the stack, to see that we really need socket: configuration entry)
But why it's using the default setting? Looks like unices are expected to connect via UNIXSocket rather than TCPSocket, and there's nothing in our connection profile about sockets, so lets add it to config/database.yml.

But how do i know this socket location? Well, on my FC6 it's pretty straightforward:


[gmarik:~/work/depot]$ netstat -an|grep mysql
unix  2      [ ACC ]     STREAM     LISTENING     36050  /var/lib/mysql/mysql.sock
unix  3      [ ]         STREAM     CONNECTED     413895 /var/lib/mysql/mysql.sock

and add


socket: /var/lib/mysql/mysql.sock

to depot_development (or depot_test|depot_production) section.

After this tinkering we get back to business, developing our depot application:)

Nice!

Duck typing makes sense!

2006-12-21T23:15:00.000+02:00

I'm reading "Programming Ruby" 2 edition, and there's a paragraph about Duck Typing, which says

You don't need to check type of your arguments

And checking objects against supported operations is just enough.

In Java, to ensure argument(object) supports certain method it's required(generally) to check ones type. But in case we just need to make sure object supports 1 single method (operation)why'd we need to check its type, which also ensures object has other methods( seems like extra info we're not interested).

So scope benefits of Duck Typing!

Duck typing allows us be as granular as possible, ensuring objects capable doing things...

And we can always check the type of course!
Yeah, duck typing makes sense!

eval.call(window, this.text) syntax error

2006-11-18T03:26:00.001+02:00

I was playing with jQuery and its jQuery Form plugin.
Everythig went fine and i got AHAH calculator working until i tried to test IE...
And yes, it didn't work...Each time a got unknown error.
So I took a debugger and started fixing...
Eventually I found the place causing error


...
call.apply(window, this.text);
...

which is to evaluate scripts from received HTML response.
But what actually caused the error was that each script was commented out like:


<script type="text/javascript">
<!-- 
alert('Im commented') 
-->
</script>

and MS IE eval function implementation isn't capable evaluating that.
So


eval("<!-- alert('Im commented')  -->"); /** Causes Syntax error in IE (6 tested)*/

while


eval("alert('Im commented')"); /** Works fine*/

Yeah, that worth to know...

Seems that Firefox behaves badly as well evaluating commented javascript, it just goes crazy without letting you know about...:)

UPDATE: well, the solution to this stuff is to comment javascript HTML comments .So eval() will not see HTML-comments:


<script> // <!--
...
//--></script>