A closer look at ruby gems

Have you ever been stuck at "Bulk updating Gem source index for http://gems.rubyforge.org/" while installing a gem?!Due the to the latest drop in the Internet connectivity in the middle east and my own country, Egypt, installing gems started to be like hell.. I had to wait for hours before having my library installed so I decided to have a closer look. surprisingly.. it was due to a bug in rubygems 1.0.1.

You can easily overcome this by updating to the latest gem 1.3.1 through gem update --system

Gem downloads libraries metadata about all available gems in a gem server. by default we have gem source "http://gems.rubyforge.org/" and recently i added "http://gems.github.com".gem saves this data in a cache file before it performs the required operation, so it doesn't have to download it again the next time.

Part of the code that reads the cache looks like this:


MARSHAL_FIELDS = { -1 => 16, 1 => 16, 2 => 16 }
#...
# Load custom marshal format, re-initializing defaults as needed
def self._load(str)
   array = Marshal.load str

   spec = Gem::Specification.new
   spec.instance_variable_set :@specification_version, array[1]

   current_version = CURRENT_SPECIFICATION_VERSION

   field_count = MARSHAL_FIELDS[spec.specification_version]

   if field_count.nil? or array.size < field_count then
      raise TypeError, "invalid Gem::Specification format #{array.inspect}"
   end
...
end

That above load method is executed to load and validate the metadata of each library from the cache unless an exception is thrown. field_count represents the expected number of attributes per gem according the that gem specification version (either -1, 1, 2 but it is 16 in all cases). the keys of the MARSHAL_FIELDS hash are numbers. but the spec.specification_version is string representation of that number. so MARSHAL_FIELDS[spec.specification_version] is always nil.

Changing it to MARSHAL_FIELDS[spec.specification_version.to_i] made the rubygems work like any civilized tool :).

Comments

Post a Comment

Popular posts from this blog

Success of Startups

Entrepreneurship has become one of the one of the most interesting topic in the business side in any country in the world. While it is digging its way in the middle east, there is some sort of confusion and a lot of aspects related to that topic. A Startup can act and advance to tackle business opportunities in a way that competes with large enterprises. and most of the times, they need to be supported by Investors to back them up. One of the things that started to really bother me in the culture is the definition of success of a startup. No one denies that a decent revenue generating startup is a successful one. However, some people believe that just raising an investment is a success.... umm, debatable. I believe we can make some kind of analogy with some entities that showed up in the past decade like biznas and skybiz (I'm sure some of you will remember those names).   they used an invalid scheme summarized as: People can register and join the network for a fee (as an i
Read more

Prime Numbers Generator

Image
did u know that 34136029 is a prime number?!! well.. that's what i recently realized :) I thought of writing a piece of code for that purpose.. wondering how far i can get. The idea of the code is simple We know that 2 and 3 are prime numbers the code checks every following odd numbers to be prime or not. To verify a prime number, the number divisibility by the prime numbers generated so far is checked. (29 is a prime number.. because it id not divisible by 2, 3, 5, 7, 11, 13, 17, 19 nor 23) We only need to verify divisibility against the numbers smaller than or equal to the root of the number in hand. (We needn't check the divisibility of 33 with 11, as we already must have found out it is divisible by 3).. I did not want to go with the mathematical root evaluation as i am not sure about its load. I used bits handling instead. the root of a number uses at most half of the number of bits that the original number uses. i only made a naive java implementation so far. downloa
Read more

Stress Testing with Siege and Bombard, know your limits

Image
I couldn't find a clear quick intro on getting siege and bombard in action, so I'm writing one here. Siege is a load testing and benchmarking utility that has been available for quite a while, It allows you to hit your web application on a specific url (or a set of urls in a file) with specific concurrency and size settings.  siege summarizes the measures of the test outcome including: Transaction rate (requests/sec) Actual concurrency (even if you hit with 200 concurrent connections, your server might be responding with just 80) Average, longest and shortest response time example: triggering 200 concurrent users, each hitting the url twice siege -c200 -r2 http://www.modsaid.com/ summary: Transactions:         400 hits Availability:      100.00 % Elapsed time:       38.44 secs Data transferred:        0.26 MB Response time:       10.09 secs Transaction rate:       10.41 trans/sec Throughput:        0.01 MB/sec Conc
Read more