ADHEAVEN - Natali Ardianto Official

Your efforts are futile (a message to DDoSer) Note:...

Fri, 18 May 2012 00:32:00 +0700

Your efforts are futile (a message to DDoSer)

Note: Month-to-month stats

Is mobile e-commerce platform dependent? As seen here in...

Sun, 06 May 2012 02:59:22 +0700

Is mobile e-commerce platform dependent? As seen here in analytics, am not too happy with the sales coming from mobile devices.

On the other hand, the infamous OMGPOP, was always working on Facebook Web Apps, and it flunk (ref). The Facebook Web Apps was only successful after Zynga acquisition.

So my premises for now: Mobile e-commerce sales will be big, IF it’s a native apps.

ps: data is taken months behind, just to give an idea

Solving problem with 502 Bad Gateway in HAProxy

Wed, 02 May 2012 03:01:00 +0700

Lately I get 502 Bad Gateway coming from HAProxy A LOT. After activating the log, you can see this message:

May 2 02:49:43 localhost haproxy[14092]: 178.255.154.2:41408 [02/May/2012:02:49:25.801] HTTP HTTP/http_02 0/0/0/-1/17463 502 21924 - - PHNN 2/2/2/1/0 0/0 “GET / HTTP/1.0”

Looking at this documentation: http://code.google.com/p/haproxy-docs/wiki/checkcache it says that I am having some kind of invalid cookie or header formats in my apps. Well I don’t care. :P So basically to stop this fiasco, all I have to do is add this at defaults:

no option checkcache

And voila! All your pain and misery has been erased.

[eAccelerator] This build of "eAccelerator" was compiled for PHP version 5.3.8. Rebuild it for your PHP version (5.3.10) or download precompiled binaries.

Tue, 01 May 2012 00:41:00 +0700

If you ever found this message:

[eAccelerator] This build of “eAccelerator” was compiled for PHP version 5.3.8. Rebuild it for your PHP version (5.3.10) or download precompiled binaries.

Then all you have to do is run the following commands (I include downloading eaccelerator just in case):

# mkdir /usr/src/eaccelerator/
# cd /usr/src/eaccelerator/
# wget -O eaccelerator-0.9.6.1.tar.bz2 http://downloads.sourceforge.net/project/eaccelerator/eaccelerator/eAccelerator%200.9.6.1/eaccelerator-0.9.6.1.tar.bz2?r=http%3A%2F%2Fsourceforge.net%2Fprojects%2Feaccelerator%2Ffiles%2Feaccelerator%2FeAccelerator%25200.9.6.1%2F&ts=1316714345&use_mirror=surfnet
# tar -xvf eaccelerator-0.9.6.1.tar.bz2
# cd eaccelerator-0.9.6.1
# phpize
# make clean
# ./configure
# make & make install

Jay Robinson: Quick note on why you should always use Progressive JPEGs if larger than 10KB

Mon, 30 Apr 2012 02:45:41 +0700

Jay Robinson: Quick note on why you should always use Progressive JPEGs if larger than 10KB:

jayrobinson:

Lately I’ve been doing a ton of reading on decreasing website load time, image optimization techniques, responsive loading for different devices, and every detail I can find in between. One quick takeaway is that I will always tick the “progressive” checkbox in Photoshop’s Save for Web dialog…

The left one is from my Gmail account (yes.. I have 13 GB...

Wed, 25 Apr 2012 03:18:19 +0700

The left one is from my Gmail account (yes.. I have 13 GB of emails!). The right one comes from Google Drive. So basically in total, I have 52 GB? Glitch or not?

60 nostalgic DOS games for 286/386/486… and I played 30 of...

Mon, 23 Apr 2012 01:59:01 +0700

60 nostalgic DOS games for 286/386/486… and I played 30 of them!

1. The Simpsons; 2. Teenage Mutant Ninja Turtle; 3. South Park;...

Fri, 20 Apr 2012 20:31:36 +0700

1. The Simpsons; 2. Teenage Mutant Ninja Turtle; 3. South Park; 4. The Smurfs; 5. Asterix & Obelisk; 6. Bert & Ernie; 7. Donald Duck and Huey Dewey Louie; 8. ?

danlewis:

I’ve figured out six of the eight. I don’t know what row 3, column 1 represents, nor row 4, column 2.

Koprol dua tahun yang lalu

Wed, 11 Apr 2012 00:18:44 +0700

Saya ingat sekali ketika kami sedang menghadiri #StartupLokal Meetup v.1 (ya! Yang pertama), saat itu Satya hadir di meetup itu. Dia pendiam sekali saat itu.

Tidak berapa lama kemudian, berita akuisisi Koprol beredar dimana-mana. Ternyata dia sedang menahan diri supaya berita tentang proses akuisisi tidak kemana-mana.

Saya pikir, Koprol menjadi awal dari domino industri startup di Indonesia.

Karena Koprol, berbondong-bondong Venture Capital datang ke Indonesia. Karena Koprol, perusahaan sebesar Yahoo! datang ke Indonesia.

Sekarang Yahoo! meng-layoff semua tim development Koprol. Apakah efeknya terhadap industri startup di Indonesia? NOTHING. Semua orang tahu bahwa alasan Yahoo! meng-layoff tim di Indonesia bukan karena mereka tidak kompeten, namun lebih karena mereka sedang mengarahkan strategi mereka sebagai perusahaan digital media, dan sebuah tim development tidak ada dalam peta mereka.

So guys, this is the ending of a beginning, and now we are writing the real chapter of the story. Good luck startup lokal!

If you want to make something similar to Silicon Valley, what...

Mon, 09 Apr 2012 03:51:15 +0700

If you want to make something similar to Silicon Valley, what should we do?

Handling user password: How do you hash your password?

Sun, 08 Apr 2012 20:08:00 +0700

Handling user password: How do you hash your password?:

After reading this article, I am starting to think that adding 34 characters (salt) in your password, and then hashing it is still ineffective.

Lets begin with hashing. This is how long does it take to brute force attack certain character length hashed text using modern days GPU:

all 6 character password MD5s: 3 seconds
all 7 character password MD5s: 4 minutes
all 8 character password MD5s: 4 hours
all 9 character password MD5s: 10 days
all 10 character password MD5s: ~625 days
all 11 character password MD5sfuggedaboudit

I have a plan. There are some techniques to add the “characters” in hashing (and make it super unique).

CONCAT (password , password, password) : By duplicating the password more than 1 times, we can harden the hash. We force user to use a minimum 6 character password, and then if we duplicate the password, then there are at least 12 characters password. Weakness: Usual suspect password, cracker create a crack pattern (if they know our technique).
CONCAT (password, salt, password): This is stronger. Unless they know our salt (if they actually hacked our server instead of only our database), this is a lot better.
CONCAT (password, salt, email): I once use password and email as a combination of hashed password, but user have to change password if they want to change the email. But I guess this is strongest, since usually email is long (username @ domain dot TLD). But of course, if they got our password hash, they usually got our user email address too.
I have seen a technique where they double hash a password. In term that they call MD5 twice. I think this is ineffective, since from the first hash to the second hash may have given collision hash structure.

How do you handle usually handle your user password? Please share with me.

salsabeela: Our 2nd anniversary was a total blast. Fantastic...

Sun, 08 Apr 2012 18:54:13 +0700

salsabeela:

Our 2nd anniversary was a total blast. Fantastic crowd, high quality speakers, generous sponsors, solid internal team! Lovin’ every bit of it

satya: List of Yahoo! Koprol developers

Sat, 07 Apr 2012 03:56:28 +0700

satya: List of Yahoo! Koprol developers:

satya:

These are software engineers, mobile developers, UI designers, program managers, quality engineers and a services engineer who worked for Yahoo! Indonesia, Yahoo! Mail, Yahoo! Messenger, Yahoo! Hub, and of course Koprol. They love Scrum, Ruby on Rails, mobile apps (J2ME, BlackBerry, Android,…

The Sounds Of Nature…Literally

Wed, 21 Mar 2012 12:36:47 +0700

The Sounds Of Nature…Literally

Installing Sphinx Search on Basic 64-bit Amazon Linux AMI

Sun, 29 Jan 2012 00:14:46 +0700

This is a simple step by step in installing Sphinx Search on Basic 64-bit Amazon Linux AMI. This is applicable to 32-bit too. This installation is very bare, starting from zero installation at all.

The conditions I wanted to meet:

Using the very latest Sphinx Search documents. And when I say latest, I mean it. I am using Sphinx Search trunk, because I need the RealTime Indexing (RT Index) and the latest trunk had the TRUNCATE RTINDEX command to remove all indexes.
Uses MySQL to connect to Sphinx Search.

So here goes:

# yum install mysql mysql-server mysql-devel
# yum install make gcc gcc-c++
# yum install svn
# cd /usr/src/
# svn checkout http://sphinxsearch.googlecode.com/svn/trunk/ sphinxsearch
# cd sphinxsearch
# ./configure —with-mysql
# make
# make install
# /usr/local/bin/indexer —all
# /etc/init.d/searchd start

And that’s it! Crazy simple!

nuniek: Peserta #startuplokal meetup termuda malam ini ^.^...

Wed, 25 Jan 2012 01:18:19 +0700

nuniek:

Peserta #startuplokal meetup termuda malam ini ^.^ (Photo by nuniek)

Project Eden, Grha 9 5th floor, Jl. Penataran no. 9,...

Thu, 19 Jan 2012 17:38:26 +0700

Project Eden, Grha 9 5th floor, Jl. Penataran no. 9, Menteng

salsabeela: Been trying to put the painting on the wall for...

Thu, 19 Jan 2012 17:23:15 +0700

salsabeela:

Been trying to put the painting on the wall for weeks. Finally after the girls (me & nuniek) spend more time at the office, the painting finally put in the right place :D

nuniek: Welcoming foreign investor to @projecteden campus this...

Thu, 19 Jan 2012 11:11:12 +0700

nuniek:

Welcoming foreign investor to @projecteden campus this morning, n they gave us good ideas :) (Photo by nuniek)

Using Sphinx as denormalized table: Need your point of view

Sat, 07 Jan 2012 00:07:00 +0700

Hey guys, I need your point of view on this. So I have this query, which can take up to 17 (Seventeen!) tables into one query. I have been using Sphinx a lot for searches, and luckily, since Sphinx 1.10-beta, we have this attribute called sql_attr_string. What this attribute does is store the value for retrieval at Sphinx. So this is what I get (below is a sphinx index):

mysql> select room_date, room_name FROM rooms LIMIT 5;
+-----------+--------+-----------+---------------+
| id        | weight | room_date | room_name     |
+-----------+--------+-----------+---------------+
| 120120106 |      1 |  20120106 | Superior Room |
| 120120107 |      1 |  20120107 | Superior Room |
| 120120108 |      1 |  20120108 | Superior Room |
| 120120109 |      1 |  20120109 | Superior Room |
| 120120110 |      1 |  20120110 | Superior Room |
+-----------+--------+-----------+-------------------+
5 rows in set (0.01 sec)

So as you can see, I don’t need to query my tables anymore, basically sphinx search is doing the hard work by indexing ALL tables (schedule in background), and updates anything necessarily by using real-time indexing.

Reading the manual, it says that searchd will cache all values in RAM. But I think that’s before 1.10-beta. I need your PoV on this. Is this plausible or is this consuming a lot of Sphinx Search resource?