Search
Jack Chen Songyong
Lightning Talk: Grow from Small Simple Steps Forward
Edit
Summarized using AI

Lightning Talk: Grow from Small Simple Steps Forward

Jack Chen Songyong • June 24, 2016 • Singapore • Lightning Talk

In Jack Chen Songyong's lightning talk at the Red Dot Ruby Conference 2016, he discusses the theme 'Grow from Small Simple Steps Forward,' using his experience with contributing to an open-source project as a case study. Jack details his journey in trying to contribute to the 'Active Record Import' library while navigating some challenging technical problems. He begins by explaining the convenience of the library, where multiple records can be imported into a database efficiently rather than through individual, redundant SQL statements.

Jack recounts the technical challenges he faced when he encountered issues with the enum feature introduced in Rails 4.1, particularly how strings and integers interact in type casting. Throughout the talk, he reflects on the confusion that arose due to the legacy code in the library and discussions with the maintainers. He shares how he learned about the inner workings of Active Record, particularly around connection adapters and type casting strategies.

The key points of his presentation include:

  • Learning from Mistakes: Jack emphasizes that contributing to open-source projects doesn't require vast experience and can begin with minor contributions.
  • Do Not Expect Perfection: He encourages newcomers to view pull requests as opportunities for learning rather than expecting them to be perfect.
  • Engagement with the Community: He highlights the importance of communication and learning through conversation with maintainers and other contributors.
  • Celebration of Small Wins: Jack stresses that while some contributions may seem trivial, they often lead to significant learning opportunities and growth.
  • Practical Knowledge Acquisition: By tackling these challenges, he gained deeper insights into Rails, database types, and the mechanics of the library he was contributing to.

In conclusion, Jack's journey illustrates that progress in open-source contributions might be gradual and complex, often requiring patience and persistence. He encourages others to contribute without compromising their engagement and to keep in mind that even small steps can lead to substantial growth in knowledge and skills. Ultimately, the experience of contributing to an open-source project can be enriching, leading to valuable connections and personal development.

Lightning Talk: Grow from Small Simple Steps Forward
Jack Chen Songyong • June 24, 2016 • Singapore • Lightning Talk

Speaker: Jack Chen Songyong, Technical Director, Skymatters Limited

Many great developers love to contribute to open source code base. But you don't have to be very experienced before doing so. This short talk is to take one of my pull requests to a Ruby gem library as an example, from the incentive, the conversations between maintainers and me, to what I learnt.

Speaker's Bio
Jack works at Skymatters Limited as the technical director. He is keen to open-source projects and gave talks at tech conferences (esp. Codeaholics HK). He also believes running a marathon and coding a project have so many in common.

Event Page: http://www.reddotrubyconf.com

Produced by Engineers.SG

Help us caption & translate this video!

http://amara.org/v/ONqK/

Red Dot Ruby Conference 2016

00:00:13.719 director at Sky matters and he's from
00:00:15.839 Hong
00:00:18.560 Kong thank
00:00:21.960 you okay uh hello everyone uh my name is
00:00:25.800 Jack today I'm going to talk about grow
00:00:28.039 from small simple steps so learning from
00:00:30.920 Amazon the name can be shorter gr from
00:00:34.040 S3 uh yeah today I'm going to walk you
00:00:37.120 through one of my pel experience
00:00:39.520 contributing to a open source project um
00:00:42.760 which was not going really well
00:00:44.960 surprisingly uh the problem is like this
00:00:47.320 the library name is called active record
00:00:49.399 import so how does it work once you need
00:00:52.719 for example import 10 books records into
00:00:55.680 a database at the same time right
00:00:58.000 instead of writing 10 different SQL
00:01:01.039 statement and 10 different uh 10
00:01:04.080 database connections you can write one
00:01:06.680 and just say book do import books and
00:01:09.360 then all the data all the record insert
00:01:11.400 in the database right uh so it helps a
00:01:13.799 lot if we need to insert millions of
00:01:15.680 data at the same
00:01:17.280 time uh yeah and but now uh some problem
00:01:21.720 comes once we use enom in rails so enum
00:01:25.439 is a feature since I think rails
00:01:28.479 4.1 um basically you can Define the inam
00:01:31.600 attributes with some uh for example the
00:01:33.840 status here must be either drafted
00:01:36.280 finished or published uh but somehow the
00:01:39.320 value save in a database uh is always
00:01:42.159 integer so so default so by so drafted
00:01:46.040 is zero finished is one and published is
00:01:50.119 two uh however to you to do the
00:01:52.600 assignment you can say uh do status
00:01:56.680 equal to zero or do status equal to jaft
00:02:00.320 uh yeah only integer is saved um
00:02:02.960 sometimes in the migration you need to
00:02:04.920 have you you want to have like default
00:02:07.039 value uh to the Staters but once we use
00:02:10.879 that again boom as argument error says
00:02:14.920 integer zero is not a valid so my first
00:02:19.560 question is um looks like I'm not I'm
00:02:22.840 not sure why the integer zero is used
00:02:25.160 instead of so why the string is used
00:02:28.239 instead of the integer then I look at uh
00:02:31.680 how does it work in the pryle because I
00:02:34.080 was using prle and find out there's a
00:02:37.160 method called PG uh get x p expr and
00:02:41.840 this is the method to get the default
00:02:43.560 value uh and the output is actually a
00:02:46.080 string so that's why we get string zero
00:02:48.159 instead of the integer and then I think
00:02:50.440 rails must be small enough to do the
00:02:53.000 type casting and we can always assign uh
00:02:56.080 integer attributes with uh string and it
00:02:59.800 it should be automatically do the
00:03:01.360 conversion um but somehow in the in the
00:03:03.840 input methods in the library uh it it
00:03:06.879 just use read attributes before type
00:03:09.120 casting uh so it basically just read
00:03:11.480 attributes from the instance without
00:03:13.280 doing any type casting uh and then uh
00:03:16.040 the place the place uh raise error is
00:03:19.720 the the the one we call method s k
00:03:22.720 equals to
00:03:23.879 V uh yeah and this is place because zero
00:03:26.840 is not a predefined uh string
00:03:30.280 and uh yeah so so there's error um and I
00:03:34.760 was also asking why the test all the
00:03:37.040 tests got to P if you still take care
00:03:39.920 about still think care of test so this
00:03:41.959 is a model file inside the spec uh it
00:03:44.959 basically defines the enum sters once
00:03:46.959 the active record version is larger than
00:03:50.760 4.1 uh for active record version 31 32 4
00:03:55.079 41 and 42 does anyone find the problem
00:03:58.040 what's wrong here
00:04:02.680 yeah the 2i because for 2i at the most
00:04:06.760 it converts the version to 4 point just
00:04:09.560 convert to four and the value is never
00:04:11.799 larger than 4.1 so the enum attribute is
00:04:15.000 never
00:04:16.160 defined okay so I get the problem I know
00:04:18.919 what's wrong I got the fail test case
00:04:21.359 now i r roll up my slaves okay work hard
00:04:25.919 overnight and then I received a lot of
00:04:29.160 encourage
00:04:30.520 ment and at the
00:04:33.800 end okay sorry I go
00:04:37.639 back I'll go back sorry
00:04:43.280 little ah so wait a second
00:04:48.560 so so yeah uh and and I reive a lot of
00:04:53.919 uh encouragement and and at last the the
00:04:57.560 author of the repo and it says it is a
00:04:59.600 best pro request I have ever seen
00:05:02.360 awesome but it didn't happen at all so
00:05:06.240 the the reality is okay the reality is I
00:05:10.560 sat down and was not sure what was wrong
00:05:13.880 the first I thought okay it might be a
00:05:16.360 problem with postr SLE but it turns out
00:05:18.960 okay s lights and myl are the same and
00:05:22.880 then I think okay that might be a
00:05:24.720 problem because rails I think rails must
00:05:26.960 be smarter than this but maybe way way
00:05:30.160 because enum it accepts strings like
00:05:32.680 draft and it never should be convert
00:05:36.400 that the string to to integer
00:05:38.720 automatically and At Last I Define okay
00:05:41.680 I I I want have a full request to
00:05:43.479 replace uh read attributes before type
00:05:45.960 cast with read attributes which
00:05:48.080 basically uh do the type casting and
00:05:51.000 then the first maintainer comes here say
00:05:54.000 okay why do uh why uh in our library
00:05:57.280 read attributes before typ casting is
00:05:59.120 used H I'm a bit confused because I
00:06:02.120 thought this is a question I should ask
00:06:04.319 him and he should told me
00:06:07.720 right uh I was bit confused and then
00:06:10.520 here comes another uh and here comes
00:06:12.880 another commuters he said oh actually
00:06:15.199 the code was back to 2007 and but but go
00:06:19.479 ahead go go to use real uh read
00:06:21.720 attributes if that got test case pass
00:06:24.240 really 2007 you know that was rails
00:06:27.759 1.2 uh I got really sad about this and
00:06:31.199 then here this comes another uh the
00:06:34.720 first commuter actually come back and
00:06:36.919 say okay actually the type cting is
00:06:39.759 handling somewhere else but uh but go
00:06:42.840 ahead to to use read attributes and I
00:06:44.960 was even confused because it's not
00:06:46.840 logical at all if there the type casting
00:06:49.120 is handled somewhere else why need to
00:06:50.960 use read attributes to do the type
00:06:53.199 casting
00:06:55.039 duplicated uh confused confused and then
00:06:58.599 uh I tell them okay okay read attributes
00:07:00.599 is actually slower for the type casting
00:07:02.319 50 times slower um because I think my
00:07:05.479 Pro request was actually not that
00:07:08.319 good uh when I was still thinking there
00:07:10.599 was another commuter actually comes uh
00:07:13.080 in in another Pool request actually made
00:07:15.360 a very simple fix and kind of get all
00:07:17.560 the Tex pass and he just basically
00:07:20.080 changed the model doent k equals V with
00:07:23.960 model square bracket k equal V uh I
00:07:26.960 thought these two are the same but
00:07:28.680 somehow there are
00:07:30.560 not uh because the square bracket equal
00:07:33.639 sign basically is uh the same with
00:07:37.360 rights underscore attributes and which
00:07:39.840 does type cussing as well I'm thinking
00:07:42.919 okay it's it's it's working but not not
00:07:45.400 very good either uh but at the moment
00:07:47.759 when everyone is was stucked uh the the
00:07:51.479 Comm the maintainer say Okay I I want to
00:07:53.960 close this I want to merge your pro
00:07:55.479 request uh go ahead but I'm going to
00:07:57.680 create another Pro request and emerge
00:07:59.919 the changes of of both of your your your
00:08:02.759 PO requests so this is uh yeah this is
00:08:07.319 kind of the the story and through the
00:08:09.560 entire story I had no much like positive
00:08:14.280 emotions uh and and the the result was
00:08:17.080 not very good my although my PO request
00:08:19.599 got merged uh but I don't have much code
00:08:22.680 contributed uh but even this I get some
00:08:25.520 takeaways and the takeaways is really
00:08:27.720 something I'm going to talk about today
00:08:29.479 so the first I know much more about the
00:08:32.000 active record connection adapter it is
00:08:34.880 good design to support your uh
00:08:37.919 application support multiple database
00:08:40.279 types and which is scalable and second I
00:08:43.599 understand more about different types of
00:08:45.040 assignment in active record and type
00:08:47.440 casting and also understand how at least
00:08:50.560 default value Works in postra sqle and
00:08:54.120 the library itself is actually pretty
00:08:55.600 good the people amazing this is a good
00:08:57.839 example uh or libr liate to uh as a
00:09:00.800 showcase how to support multiple ruby
00:09:03.120 versions and database adapters at the
00:09:05.240 same time and of course no uh reals you
00:09:08.600 know more and and here's what are
00:09:10.680 takeaways I come here to for you so the
00:09:13.560 first don't believe in Stars so for this
00:09:15.920 Library it has more
00:09:17.839 than so for this Library it has more
00:09:20.240 than 1,500 stars and it is updated very
00:09:23.399 frequently but even for this there was
00:09:25.760 still some mistakes um I think most of
00:09:28.399 the beginners can
00:09:29.959 fix and then uh sorry second don't think
00:09:33.160 of making a p a perfect pool request at
00:09:35.480 the first so learn from the
00:09:37.240 conversations from others uh and last I
00:09:40.399 think contributing to open source
00:09:42.240 project uh sometime is Trivial like
00:09:45.360 small simple steps but sometimes even
00:09:48.000 the small simple steps is not that easy
00:09:50.839 is sometimes not
00:09:52.399 smooth sometimes the reality is like
00:09:55.640 this so my so so my first my my sorry my
00:09:59.360 last suggestion is like go ahead don't
00:10:02.120 compromise thank
00:10:06.640 you
JCS
Jack Chen Songyong
1 talk
Explore all talks recorded at Red Dot Ruby Conference 2016
+17