Search
RubyConf 2011
Ruby Conf 2011 MacRuby on Rails
Edit
#ruby-on-rails
#objective-c
#garbage-collection
#ruby-internals
#desktop-applications
Summarized using AI

Ruby Conf 2011 MacRuby on Rails

Kouji Takao • September 29, 2011 • New Orleans, Louisiana • Talk

The video titled "Ruby Conf 2011 MacRuby on Rails" features a presentation by Kouji Takao on the integration of MacRuby with Ruby on Rails, focusing on his modifications to make this compatibility possible. MacRuby, an implementation of Ruby 1.9 utilizing Mac OS X technologies, originally lacked the functionality required to run Ruby on Rails applications.

In the presentation, Takao outlines the following key points:

  • Introduction of MacRuby: MacRuby is designed to be compatible with Ruby 1.9 both syntactically and behaviorally. Takao gives a brief overview of the building blocks of both Ruby 1.9 and MacRuby highlighting how their structures differ.

  • Development Challenges: Initially, Takao encountered many challenges while trying to install and run Ruby on Rails on MacRuby. He describes his five-step problem-solving approach: learn the Rails framework, identify where failures occur, research the causes, develop patches to fix the issues, and verify these fixes against Ruby specifications.

  • Example Issues and Fixes: He shares specific examples of issues he fixed:

    • Exception Handling: Takao addressed problem number 860 related to catch and throw exception handling by modifying the way exceptions are processed in MacRuby, ensuring compatibility with Rails.
    • Constant Lookup: Takao tackled issue number 11192, which involved constant lookup failures. This included modifying the behavior of constant resolution within modules and classes to comply with Ruby specifications, which was crucial for Rails' functionality.

  • Demonstration: Towards the end, Takao successfully demonstrates creating a new Rails project, generating scaffolding, and starting the server, showcasing the progress made. However, he also acknowledges that further work is needed to fully realize Ruby on Rails functionality in MacRuby.

  • Conclusion: The presentation emphasizes Takao's dedication to improving MacRuby's capabilities and reveals the ongoing challenges developers may face with such integrations. He expresses gratitude to those who supported his journey in this technical endeavor, illustrating both the complexities and the excitement surrounding development in the Ruby ecosystem.

Overall, Kouji Takao's insights on modifying MacRuby highlight an important phase in bringing Ruby on Rails support to Mac environments, along with the technical hurdles that developers may face.

Ruby Conf 2011 MacRuby on Rails
Kouji Takao • New Orleans, Louisiana • Talk

Date: September 29, 2011
Published: December 12, 2011
Announced: unknown

MacRuby is an implementation of Ruby 1.9 that is directly on top of Mac OS X core technologies. Recently, MacRuby has become viable as a tool for developing useful desktop applications for Mac OS X. However, as of March 2011, MacRuby is still missing some functionality that is present in cRuby. Therefore, MacRuby is not able to run Ruby on Rails. In my presentation, I will explain how I modified MacRuby to make it a suitable foundation for running Rails. I would also like to explain some of technical intricacies that I discovered along the way.

RubyConf 2011

00:00:17.160 I am about to start my presentation I would like to welcome all
00:00:23.279 of you it is a great honor for me to be here
00:00:30.960 please let me introduce myself first and let me introduce my
00:00:37.680 company my name is Ki taka both kji and taka are commonly used
00:00:46.039 a first name in Japan in my case Ki is
00:00:51.680 my first name and taka is my last name
00:00:57.320 please call me coji I'm a research engineer at Network
00:01:05.880 applied communication laboratory LTD the Japanese company that hire mat and
00:01:14.720 others as for me my work includes anything related to system integration
00:01:22.759 for example system design programming planning and meeting with the customers
00:01:29.520 and I will work are the leader of small
00:01:34.759 team I develop open source software during my spare
00:01:39.840 time I'm A mby and C committer in maby I am in charge of some
00:01:47.560 of the cor Parts in C I am in charge of the RM
00:01:56.880 mod you can contact me on email Twitter or
00:02:05.079 Facebook as I said previous previously my company's name is Network applied
00:02:12.080 communication laboratory LTD my company is also known as
00:02:21.519 n as you probably know N means salt salt
00:02:26.640 is necessary for all creature to live this this is why we Nam the company
00:02:36.760 n you can check out n homepage at www.
00:02:43.000 njp but almost all contents are in
00:02:48.760 Japanese NL is located in shiman prefecture in
00:02:55.560 Japan the company was founded in 200 and
00:03:00.680 one so 2011 is the 10th
00:03:06.480 anniversary by the way it's also my own 10th anniversary as an employee I'm very
00:03:16.959 happy oh thank you uh uh uh 15 uh uh 57 people uh work at
00:03:28.480 my company woo that's a lot of rubies is just
00:03:36.280 joking oh actually about half of my co-workers work using Ruby and six
00:03:43.040 people are C be commits the company works with system
00:03:50.000 integration training and so on please let me introduce the N way as
00:03:58.439 I see it this is our president his name is hirosi
00:04:06.879 in in a speech at my coworker wedding party he said to my
00:04:14.360 coworker first I want you to be
00:04:19.959 happy next I want you to make your family
00:04:25.759 happy or finally if you have more happiness like then please share it with our
00:04:36.880 company so I think the priority of my company is as
00:04:42.639 fors first the employees will being next the wellbeing of the Empire's
00:04:50.840 family and finally the company itself so
00:04:56.840 this is why I am my company yeah now let's continue on to the main
00:05:05.160 topic of today macro on
00:05:12.080 RS in my presentation I will explain how I modified maby to make it a suitable
00:05:19.479 foundation for running RS I would also like to explain some of the technical
00:05:25.360 internals that I discovered along the way
00:05:34.560 first I would like to First brief explain about
00:05:42.360 mroe mroe is a unique brand of Ruby 1.9 and Objective
00:05:50.639 C the goal of the M project is to be 100% compatible syntactically and
00:05:58.960 behaviorally with Ruby 1.9 oh this is Ruby
00:06:06.599 1.9 this system diagram shows the main building blocks of Ruby 1.9 the top side
00:06:13.880 shows the Brock blocks facing the user and the bottom side shows the BR facing
00:06:20.759 the hardware at the at the top we have the
00:06:26.560 applications made by the Ruby user below that we have Ruby 1.9 these are
00:06:33.800 Ruby's main building blocks we have Ruby BM Ruby B machine
00:06:41.080 which was previously called Yi we have the building garbage
00:06:46.800 corlection which uses Mar and seep then we have the building libraries
00:06:53.639 which includes the string class and the array class and so on
00:07:04.039 and then we have the standard rary which includes we Line open SSL and other
00:07:11.319 useful libraries below below Ruby we have the
00:07:17.160 system Ries and below the rary we have the operating system R 1.9 operates on a
00:07:25.520 number of different platforms among other Reax window do MOS stand and so on
00:07:33.080 so this is the structure of Ruby
00:07:39.639 1.9 so let's try to compare this with the system diagram as it looks for micro
00:07:46.759 where look at each element in turn the application stays as it is it
00:07:53.520 is naturally Ruby 1.9 becomes micro Ruby BM is replaced by
00:08:03.639 lbm the building garbage corlection is replaced by a mechanism provided by m
00:08:11.400 this m feature is called or zone or zone is a generation and threaded garbage
00:08:18.840 correction mechanism part of the Bing library is
00:08:24.280 replaced by a future called Foundation which is provided by Mark a
00:08:30.639 object array string hash and others are replaced this way the result is that
00:08:38.279 these classes get not only the functional functionality provided by Ruby but also the functionality provided
00:08:46.279 by foundation for instance a foundation method called
00:08:52.680 transform is add to the Ruby strength to class this method can convert the out
00:08:59.200 ver of my name kji into the corresponding Japanese hilana
00:09:08.440 characters the standard rers are more oress similar to Ruby
00:09:15.720 1.9 the system R are also unchanged and the OS is of course or Mar
00:09:24.440 Austin actually ma Ruby only runs on Mar atin
00:09:33.480 there so if we look at the whole well then it looks like this so now I have
00:09:41.160 done with my introductory explain of micro so let's continue with the main
00:09:49.000 topic M only so I'm going to talk about
00:09:54.120 making Ruby on R works or microbe
00:10:01.000 let's see what we can currently do with mcro on Riles creating a new red
00:10:09.920 project Genting a scaffold starting up the server and
00:10:16.680 showing the wecome page so it is currently possible to run a red application on micro without modified
00:10:24.880 modifying Le however it wasn't always like this oh
00:10:31.360 let's travel back in
00:10:36.639 time oh 7 months ago March
00:10:42.800 2011 in Japan I saw does rails work on
00:10:51.000 macbe I tried to install rails on MBE and I could
00:10:57.560 install so I saw okay it's install but I
00:11:03.480 didn't expect the red generators would work I tried to create a red Project
00:11:10.760 without much expectation and I
00:11:18.880 could not be my own eyes I tried to gener a
00:11:24.720 scaff then it fail and starting up the sub fa to after
00:11:32.519 all R didn't work on
00:11:38.040 micro so what did I do my Approach was to follow the five-step process showing
00:11:44.959 here repeat repeatedly first I learn ra as a next
00:11:53.639 step I see where it fails for instance segmentation Force about the programs
00:12:00.279 and so on the thir step is to research the cause of the
00:12:06.839 fair the first step is to c a patch that fixes Pro
00:12:12.399 problem and finally I Lear Ru Ruby spec to verify my
00:12:18.720 patch and repeat from the top if
00:12:25.480 necessary in uh in order to do this I need to know
00:12:31.360 about CB Objective C C++ and er VM when you just work with CB then you
00:12:41.040 only need to know about CB but when you want to code pches for M then you also
00:12:49.600 need to know about Objective C C++ and a
00:12:55.040 BM so I think that to code pches for mic is more difficult than to do so for
00:13:05.040 C now I like to discuss a couple of issues that I that I fixed in M for
00:13:16.760 examples let's have a look at the first case which is track issue number
00:13:24.160 860 and issue with catch and throw and EXP handling in
00:13:31.320 microbe this error can be reproduced by
00:13:38.120 Disco If you can run disc then the following error will
00:13:44.880 occur this kind of exception handling is used by Band so in order to make red
00:13:52.880 work we need to fix this error let me try to explain this reason
00:14:00.880 first we need to know a little bit about how exhibtion handling is done in
00:14:08.240 MCR have a look at Discord in this program we read and rescue twice let's
00:14:14.839 look at what happens when we learn this pro pro program on
00:14:20.480 MCR on line two we let exception
00:14:26.199 a or mro let exceptions are start in our
00:14:31.399 stack we call the stack the exception
00:14:39.920 stack on righty we rescue exception a and continue execution on line five
00:14:48.240 inside the rescue block we l extion b so exception B is pushed onto the stock
00:14:59.639 on line six we rescue extion B and continue ex executing the
00:15:06.360 cord on line seven we Le the end of the inner resue block so M pops one
00:15:13.759 exception of the stack and on line eight we lead the end
00:15:22.800 of the outer rescue block and again we pop an exception from from the St
00:15:29.759 exception stock so now we are done handling all all the exceptions read by
00:15:36.639 this program and the exception stack is empty this is a way that mro starts and
00:15:44.560 processes
00:15:51.759 exceptions so let's look at how exception handling with car out when it
00:16:03.040 throw have a look at this go in this Pro program we Define a catch block leled
00:16:09.160 full inside the catch block we have a begin rescue where we let an
00:16:15.800 exception and in the rescue block we use throw to jump out of the beginning
00:16:22.720 rescue let's step through the ex execution again to see what happened in
00:16:29.399 the microb exception stock just like we did in the last
00:16:35.040 example on L one we Define a rid block using catch after that if we call thr
00:16:42.040 with symbol who as an argument then we jump out of the
00:16:48.440 block on line three we let exception a so it is pushed onto the
00:16:57.079 St on on line five we call row with symol who
00:17:05.079 as argument so we jump out of the oh
00:17:10.240 sorry so we jump out of the rabbit broke the mro pops exception a from the
00:17:19.360 stock so this is how through house R takes out exceptions from the stock in m
00:17:32.360 before I patch it macr thr was implemented as you can see here this
00:17:38.840 code is from bm. CPP thr is implemented in
00:17:46.160 C++ the current exception function uh return the topmost exception from the
00:17:53.240 stock and the pop current exception function or pops one exception of the
00:18:02.799 stock now have a look at the code that reproduce the error once
00:18:09.280 again on line two we l exception a so exception a is pushed onto the
00:18:17.039 stck L four we call catch and throw then Mar pops one exception of the
00:18:24.440 stock and the stock is empty line five we Le the end of the out
00:18:32.320 rescue block and again we pop an exception of the stock however the stock
00:18:39.880 is empty therefore M ABS as you can
00:18:45.960 probably see that the cause of the ER of the eror has popped one exception of the
00:18:51.919 stack on R for let me explain how to fix the error
00:19:01.039 first let's look at what happens when we run this program program on macb after
00:19:07.760 we fix the error up until L three things work as
00:19:15.520 they always did however now things work a bit
00:19:22.880 different from line four now when we call catch then we
00:19:28.320 start a record of what is at the top of the exception s in this case exception
00:19:36.559 a next when we call through we compare the exception that we St with the
00:19:44.120 exception at the top of the exception stock
00:19:49.159 if these two exceptions are the same then mcrob doesn't do
00:19:56.320 anything if they are different then mro pops one exception of the sack
00:20:02.840 now in this case they are the same so Mac doesn't do
00:20:09.600 anything on line five MC pops an exception from the stock and then the
00:20:16.520 stack is empty so this is how the new implementation of throw
00:20:26.000 works so I change the implementation so that microb only pops exceptions when
00:20:34.240 needed so the code in the green Fram is where we compare the St exception with
00:20:41.960 the exception at the top of the stock and in this part M pops one
00:20:48.640 exception of the S so in short we keep popping exceptions of the sack until the
00:20:56.080 hard condition turn for or in other words until we reach the
00:21:02.960 recorded exception finally let's look at what
00:21:08.360 happens when we run a more complex program on macrobe with the fixed
00:21:14.120 implementation of St line two we read exception a so
00:21:22.559 exception a is pushed onto the stck right four we call catch and start exception a
00:21:30.840 line six we let extion B and push it onto the stock line eight we call row
00:21:39.320 then we compare the recorded exception a with exception B of the
00:21:44.480 St this a different so m b pops exception B of the stock then we compare
00:21:51.480 the recorded exception a with the new top of the stock which is now exception
00:21:56.679 a the are say so M doesn't do anything then we jump out of the r catch
00:22:06.760 block on line 11 M pops one exception of the stock and the pr program finishes
00:22:14.600 without aboring so this is how I modified the
00:22:24.000 program next let's look at another example of how I mod how I fix micro
00:22:31.919 this time we look at track issue number 11 192 which is related constant look up
00:22:39.000 in micro the issue was the micro fail to correct correctly look up constants in a
00:22:46.440 number of station when you use module era and class
00:22:53.480 ER this error can be repr reproduced by this disc code
00:23:00.679 if you run this SC then the following error will
00:23:05.799 occur please look at the line with object do class A in CB when we look for
00:23:13.320 the constant B we start looking within the green box so we are able to find constant
00:23:24.279 B however in the micro at the time we only looked within class a reer object
00:23:32.960 and since constant B isn't defined there the constant look up would fail in
00:23:39.880 micro constant lookup within a crass e or a modu e argument is widely used by
00:23:46.799 active support so in order to make R works we need to fix this
00:23:56.360 error to really understand on this issue you must have a detailed understanding
00:24:02.000 of how constant r c Works in Ruby and this is one of the parts of
00:24:08.679 Ruby's language specication that quiet hard to
00:24:14.279 understand and in most cases you can get along quite fine without knowing the
00:24:21.720 details of constant lookup in Ruby let's have a look at this call
00:24:27.880 where look look at it from the top and down line one we begin the defining of
00:24:35.720 modul and line two We Begin the def definition of module
00:24:43.000 b line three we Define a constant called Con in modu
00:24:49.520 b line four we open the top module a in other words the module a that we Define
00:24:56.799 in line one and finally on line five we print the
00:25:02.320 constant uh content of the constant con through the standard
00:25:07.880 output in this example CB would look for the constant con within the gra frame
00:25:15.360 green frame we call the scope represented by the green frame for elri
00:25:21.399 scope and cdb will find the constant const that is defined in module B
00:25:28.760 so you see the Conant lookup resolution is done in Le Scopes we call it Le
00:25:36.679 constant R up
00:25:43.799 Hereafter at the time maku had not implemented electrical constant
00:25:50.399 rup in this Cod on rine five maku only looks for the constant con Within modu
00:25:59.840 a therefore con cannot be
00:26:06.600 found next we have example with class and
00:26:12.200 C constru becomes even more complex when we use Class
00:26:18.919 A the behavior changes depending on whether we pass a block or a string as
00:26:25.240 Argent to class ER in the code example on the left hand we pass a block and in
00:26:33.240 the right hand example we pass a string except for the different
00:26:39.080 arguments the two example are identical first I'll explain about
00:26:46.520 coring class with a broke let me explain this code sequency from the top line one
00:26:54.279 we Define module a line two we Define the con con con line three we Define method
00:27:02.440 F line four we pass a block to class E Line five we print the Conant content
00:27:10.559 of con to the standard output line six we close the class a
00:27:17.039 block so when we call method F then line four to six we get
00:27:22.880 executed right then we Define class ke line 11 we Define another constant
00:27:31.440 called conster but this time inside Class
00:27:37.120 K as you can see on line 12 we extended
00:27:42.360 Class K with module a online St we call
00:27:48.320 module a f method as a as a result the
00:27:53.440 code in line four to six is called so can you tell home what scope the
00:28:00.559 constant Ro C starts in this
00:28:06.000 case in this case it is easy to tell we look for the constant con within the
00:28:12.120 green frame so in this case the P output the
00:28:17.519 string is
00:28:22.600 constant next let's look at performing cross EV with a storing argument
00:28:28.640 the code example only differs on line four to six line four we pass a string to class
00:28:37.120 E in the form of a here document on line five we print the con
00:28:44.320 content of const on line six we close the here
00:28:51.840 document just as before we call module as F method online set
00:28:59.399 so can you tell her what scope the constant uh lookup starts in this
00:29:10.159 case actually we would look for con within this green
00:29:16.120 frame let me explain the reason whenever we call module as ISX method we pass and
00:29:25.399 evalate the string that is passed or P to class E and the scope in which we
00:29:33.000 look for con is deter determined uh to be the scope in
00:29:39.880 which the method f is run therefore the lookup scope becomes
00:29:46.120 the content of Class K as you as you may know we call this kind of scope for
00:29:53.320 dynamic scope so in this in this case the the P
00:29:58.600 output the strength Case Case constant so let's look at the two cases
00:30:07.720 together we can see that the scope of the constant look G differs greatly
00:30:15.000 depending on the kind of argument passed to class
00:30:23.960 E it was not easy to fix this error I I had to start over twice and I was
00:30:31.519 finally able to find the satisfy Sol solution in the South
00:30:41.840 attempt I edit more than 3,000 lines of
00:30:47.600 call when I finally managed to fix this error Laurent who is the main developer
00:30:54.919 of macby wrot the following to to
00:31:00.120 me it's officially Ki is the hero of the day week
00:31:06.200 month his patches seem to fix remain lric con bags and that made me very
00:31:21.360 yeah however after enjoying a brief Victory I
00:31:26.760 soon found a new and difficult issue related to constant look up this error
00:31:33.799 can be reproduced by this score let me explain this code section from the top
00:31:39.880 line one we Define module a line two we Define con B and module a line five we
00:31:47.120 specify module a as receiver and core class ER with the do block
00:31:53.480 n line six we are inside the cross EV block so the current receiver is module
00:32:00.159 a therefore this means that we are defined the class method F or module
00:32:07.320 a line seven we out output the con constant B to the standard
00:32:15.799 output and on line 11 we call method F or module a this means that we are going
00:32:23.120 to have a look up constant B
00:32:28.480 we are now using a patched M which implements re scope this means that we
00:32:35.120 are going to start looking for constant B within the green frame naturally we
00:32:40.919 fail to find B within the green frame so we want to continue our search for B this time we
00:32:49.679 look in object for constant B this also means that we are going to look for a
00:32:56.360 constant constant cor B that is defined at top
00:33:03.399 R here I would like to add a word of caution Class E does not add its own
00:33:11.919 receiver to the constant lookup scope and this behavior is part of R's spef
00:33:19.399 specification so I expected that the search for constant B would fa
00:33:29.919 surprisingly C 1.9 F uh constant B in
00:33:36.080 module a so it succeeded to find the constant B even though it it should fail
00:33:43.519 according to the Ruby specification on the other hand Ruby 1
00:33:51.399 8.7 work as expected and F to find the constant B MBE pred as well so I asked
00:34:01.279 mat for advice about this
00:34:06.880 issue he said this is a b in CV
00:34:13.040 1.9 and soon therefore thereafter C 1.9
00:34:18.919 Behavior was fixed to match the behavior of Ruby 1.8 8.7
00:34:29.399 okay it's time to show you the result of my work making R works on mroe I will
00:34:37.879 demonstration Ruby on red learning on mroe I am using red
00:34:44.359 3.0.7 which is a bit old
00:34:58.480 first we are going to check makui and Red's
00:35:21.760 version okay uh m is 0.11 where is is
00:35:35.000 ra uh sorry uh we are going to create a new R
00:35:48.839 project please wait a bit
00:35:58.880 okay uh this is also
00:36:10.240 okay next we are going to generate a SCA
00:36:24.720 for please wait more
00:36:43.400 maybe my machine is
00:36:52.640 old uh it's a secret
00:37:04.319 this is related another B okay uh this is also okay next we are
00:37:13.240 going to start up the server
00:37:28.560 please vit
00:37:49.040 or okay
00:38:00.200 oh this is also okay finally we are going to show the welcome
00:38:25.280 page oh this is also okay this is cool isn't
00:39:16.400 up oh oh my God that's
00:39:25.079 happen iir uh whom this is
00:39:39.760 areed actually the S we about if you click this link on the welcome
00:39:57.839 making red work on M is more difficult than I expected perhaps even more
00:40:05.119 difficult than implementing RAR constant
00:40:11.839 R so the M uh train trip continues everyone please get on
00:40:19.680 board this is a URL for the M prodject
00:40:24.720 homepage this is M off Twitter account mro is on
00:40:34.400 GitHub this is the maku mailing list address so I would like to conclus
00:40:42.599 conclude this presentation with a quick summary of what I had said so
00:40:48.920 far m is a unique brand of Ruby 1.9 and Objective C the goal is to be one 100%
00:40:58.480 compatible with rup 1.9 mro only Les we are now able to show
00:41:05.079 the defa R will compage but there is still a lot of to
00:41:11.960 do L of all I would like to express my gr
00:41:18.440 gratitude toward the ruik K host and organ organizers for giving me the
00:41:25.880 opportunity to give give this presentation and I would like to thank
00:41:32.040 my coworker to Yamamoto s who helped me
00:41:37.319 with this presentation and finally I would like to thank my everyone here for
00:41:43.920 patent listening to this presentation despite my limited English thank
00:41:52.079 you thank you very much
Kouji Takao
@takaokouji
Explore all talks recorded at RubyConf 2011
+55