00:00:15.519
uh thank you for introduction so uh today I we held the Ruby hack challenge
00:00:22.320
so Ruby hack challenge is a uh the the hacking the MRI so Luby M Luby
00:00:31.800
interpr interpreter or implementation so little in see so maybe
00:00:38.879
in at the other rooms maybe they loot Ruby but in this room you need to like C
00:00:47.879
C language so maybe maybe most of people
00:00:53.399
using Luby interpreter so the small letter small four letters Ruby command
00:00:59.239
so Ruby command is uh L in C and maybe
00:01:04.720
you are using and you have some interest about that the this in software so I
00:01:11.880
want to we want to guide to hack The Interpreter
00:01:17.880
itself so the goal of this Workshop is
00:01:24.920
hack uh Luby interpreter with your faite modification or try recent features so
00:01:33.720
Ruby 3.4 will be released soon and the Ruby master branch has many new features
00:01:41.280
so please try on your laptop such a new
00:01:47.079
features and if you have any issues or
00:01:52.320
any requests so maybe M will be this a
00:01:58.560
maybe one uh later so you can discuss so
00:02:04.079
I don't like this feature or something like so you can ask him directory and the uh the quiz will be
00:02:14.280
more easier I think anyway so uh I want to I want to show
00:02:20.599
some the ground of the Luby interpreter so Luby interpreter is the B so usually
00:02:29.040
your application excuse
00:02:43.400
me usually you are liting application with many gems so such as Lu l or
00:02:50.599
something like that and the Luby interpreter is based based at the lowest
00:02:57.440
lower uh lowest level of the the software stock however the from the
00:03:04.480
Ruby interpreter perspective all of Ruby James your applications
00:03:10.440
Ru softwares or something like that everything are Ruby script and they are
00:03:17.239
pass to the abstract syntax 3 and compile it to the Ruby bite code and
00:03:23.239
Ruby uh interpret by the the interpreters so
00:03:30.680
in this Workshop uh we will uh modify this area so Li interpret
00:03:41.040
internals so we have some steps so uh we we
00:03:49.280
prepare uh uh the document maybe you can sh you already know the website about
00:03:56.400
the Ruby hack Challenge and you can read every every document so the at first the first exercise is to
00:04:06.159
change the version string so Ruby Ruby Das Bui shows the Ruby version string so
00:04:13.599
you can modify anything you want for example your name is Ruby
00:04:19.759
or python or power or something like that so anything you can you can change
00:04:26.400
the verion string maybe it is very easy step but better it's your own
00:04:32.720
interpreter so please try make to make your own Lu
00:04:38.560
interpreter and after that so please enjoy the leing the documents and try
00:04:45.080
the exercise in the documents or if you have any interest uh about to introduce
00:04:52.720
a new features such as a make a new method or making a new mechanism for
00:04:59.360
example in virtual machine garbage corlection or something like that maybe there are many experts in this room so
00:05:06.280
you can ask then to how to how to make it or something or or uh try the newer
00:05:15.479
newest Ruby interpreter so the priorities of Ruby 3.4 and so
00:05:30.720
idea so if you have the the favorite features so there is no problem so
00:05:37.800
please try to implement that or fix a bug so you can find the bug report or
00:05:45.680
feature request reports on the uh this website so bugs L all issues so many
00:05:55.160
people report to the uh new issues on this side so please check this one and
00:06:03.520
uh check check the uh feature request or bug report and for example so on The Bu report so
00:06:12.319
reproducing process is very important for us so for example so checking the uh
00:06:20.240
the reproducible code or something like that and the uh report new report so I
00:06:27.120
can also uh reproduce this issue or something like
00:06:32.160
that uh another idea is another strange idea is uh so for example we have a here
00:06:40.639
documents in Luby so here document we can write uh any text so we can write
00:06:48.160
the C source code R source code or python source code anything anything and
00:06:54.759
how about to run the here document with the for example the uh
00:07:02.120
python interpreter so this is one uh simple idea or another weird idea is to
00:07:09.039
sound the high tone at the if the so
00:07:15.759
modify the vir machine and the sound uh sound High tone when the if statement
00:07:24.800
the condition was true p and
00:07:30.759
low tone B at the condition was forse so
00:07:37.720
the program will uh sound strange music or uh so this is where idea but
00:07:46.360
the so this idea is from the students so high school students and they they can
00:07:53.159
uh sound great music with some applications or uh we so for example if
00:08:00.000
you have interest about uh about uh uh
00:08:06.120
mechanism of v machine so visualize the internal stack machine Stacks oh sorry v
00:08:12.280
machine stacks and shows the many stack
00:08:17.800
growing and growing and so on is a very
00:08:23.400
uh uh easy to see the the what the B Machine interpret that is
00:08:32.000
doing so this is only idea so you can you can make your own
00:08:39.240
idea so lecture materials is here and the schedule is here so you
00:08:47.839
can so this guide only uh 10 minutes and
00:08:55.000
after that so you please enjoy your Ruby uh hacking sorry Ruby hacking Ruby
00:09:03.519
interpret hacking on yourself and please ask us for uh if you have any
00:09:12.800
trouble of course this uh this room only
00:09:18.040
has a few tables so of course you can go out
00:09:23.680
to hug there with your laptop uh outside of this room or in your uh hotel room or
00:09:33.720
anywhere and the the I want to emphasize that uh from
00:09:39.560
the TW 12 at noon so we have uh we have
00:09:44.720
make we will make ask Ruby core developer session so please raise your
00:09:50.720
hand you you Ruby core developer so so many Ruby developers so
00:09:58.640
uh thank you so much so they can stand or sit down around there so
00:10:05.880
you can ask anything uh with uh uh to
00:10:12.640
us so after that ah
00:10:18.279
so I want to ask that to prepare this uh ask uh am session so please write your
00:10:27.040
question on the this H MD page your question to this uh page so
00:10:33.680
any anyone can write the the hugd page so please uh play please write
00:10:42.160
your questions and we can manage these questions and after lunch so please uh
00:10:50.839
continue your uh interpret hacking and after that so at the uh
00:10:59.360
last of this Workshop so maybe someone have your achievements so
00:11:07.000
please please talk your achievement at this
00:11:13.440
stage okay this is uh everything about this Workshop
00:11:20.399
so again could you so Rubik Comm please come here and
00:11:28.959
uh could you introduce yourself with one
00:11:41.519
message so you so you can you can uh
00:11:46.959
memorize the name the PA and
00:11:53.000
name this is my face my name is Eileen uh or Eileen
00:11:58.920
codes on the internet so I don't know you
00:12:05.120
next hi my name's Aaron um I like I like I like hacking on Ruby so if
00:12:12.560
you have any questions about hacking on Ruby core stuff please come say hello and I will help
00:12:20.240
you uh I'm John Hawthorne or Jay Hawthorne looking forward to
00:12:28.000
hacking uh I'm Peter zoo I work at Shopify on
00:12:33.720
Ruby GC yeah specifically
00:12:39.199
GC I'm Allan I work on wedget stuff at Shopify I'm friendly but maybe I lack
00:12:47.199
interpersonal skills compared to some of the rest of these guys so don't read too
00:12:52.279
much into my behavior and and come come ask me if you have any questions
00:13:02.279
hi I'm from Japan I I I'm a rails guy actually but I run a conference in Japan
00:13:10.160
called Ruby Ki I hope to see you all there in Japan next year thank
00:13:18.279
you I'm yke uh my uh account name is
00:13:23.560
Mame M and you may uh see my name in the
00:13:29.320
B tra of duby and so I think I have the uh the least
00:13:36.079
communication skill because I am very I am very poor in English but I will give
00:13:42.079
it TR so please uh ask anything thank
00:13:52.720
you thank you so much so the Okay so the guidance is over so
00:14:03.279
please enjoy your uh interpret hacking
00:14:14.519
enjoy uh please join the uh this slug channel on the this
00:14:21.560
conference Workshop Ruby hug CH oh yeah we don't
00:14:26.920
need all right let's do this I
00:14:33.440
guess all right Ruby core team please come up here to the stage we're going to
00:14:39.240
do Q&A now this is so this is your opportunity
00:14:46.959
to ask the Ruby core team questions air
00:14:53.720
grievances maybe get Matts to commit your favorite feature to the language
00:15:04.720
so we've got um I don't know if you all saw the the hackmd thing in the in the slack uh but we
00:15:12.480
have um we have a hackmd where you can add add some stuff to it we're just going to go through all these and
00:15:17.880
discuss discuss them uh this is a bit of a like this particular format here is a
00:15:24.160
bit of a tradition that we have at uh Ruby kige and we're bringing it here to ruon so only in a little bit smaller
00:15:31.839
format there are a few there are a few more Ruby core team members in Japan
00:15:37.600
than there are here a few just a few all right um let's see so we did we did the
00:15:44.199
hack challenge we're going to do some stuff here so we got a question our first question is I'm put it I put them
00:15:52.040
up here on the board but I I'm going to read it aloud for you all uh Matts you
00:15:57.160
published that Ruby 4 will be shipped with annotation features at Euro
00:16:04.199
2024 could you could you elaborate more kichi is asking what what is do we have
00:16:10.399
another mic ah thank you
00:16:16.199
so matx what is The annotation what is that what does that mean annotation
00:16:21.880
feature yeah I'll try to kill some more time
00:16:27.880
while we wait for the it's okay
00:16:36.279
testing yeah okay great I have another one too that yeah sure please great I'll
00:16:41.920
be right back the The annotation is kind of the the the adding meta information to the
00:16:49.600
code like a you know it's kind
00:16:54.839
of kind of similar to the for example the Cs the attribute or the the python
00:17:01.839
declaration or something like that but the the python declaration is to close
00:17:07.520
to the code so the the think about the Cs that attribute is the kind of close
00:17:13.000
thing so the uh The Meta programming code can uh retrieve the meta
00:17:18.919
information uh added by The annotation so that uh we uh add the modify the code
00:17:27.480
according to the the Val of the attribute that's but uh I have no
00:17:32.919
concrete plan yet so that it's just a vague idea so what does The annotation
00:17:38.240
look like is it going to be part of the language like
00:17:49.919
like that I haven't decided decided the any appearance yet for example the uh in
00:17:56.280
some language we uh they use the the double square bracket but it it has
00:18:11.919
open where do I put it
00:18:17.159
yeah yeah double bracket double brackets before the death before the death ah
00:18:23.960
okay okay okay so like um uh square bracket square bracket
00:18:30.320
to open open too open ah okay I see sorry like that then
00:18:39.400
something yeah or say read only or something ah okay
00:18:45.880
okay like yeah like that but double braet itself has the the
00:18:54.000
you know to proper meaning in Ruby already so that we we cannot use that but but kind of like that okay
00:19:03.799
okay you don't show The annotation talk at this R keynote yeah because it's too
00:19:12.640
big to explain in the public I
00:19:18.400
see okay uh usually I put some kind of the you know the very early ideas into
00:19:24.960
the keynote traditionally so that uh when
00:19:33.240
I UNC counted the the ideas in the the Recon Keynotes it was the uh only 70% of
00:19:42.640
the these vague ideas are implemented in the that Canon Ruby okay I'm not sure
00:19:49.200
The annotation will be which which category you know not implemented or
00:19:57.400
implemented okay should we move on to our next
00:20:03.120
question Matt's feature 2025 was merged uh in the other hands there are
00:20:09.400
some reports which show many warnings do you want to continue it what is 2025 let's take a
00:20:22.360
there we go Frozen string literal by ah I'll
00:20:37.640
this because go of the performance yes and I like it and I and I don't I hate
00:20:43.799
putting that comment because we put the comment we have to put that comment
00:20:50.520
everywhere and too many people like copy like files and like don't actually
00:20:56.760
fix stuff so it's all messed up they don't know what they're doing so just get rid of that comment I hate it yeah
00:21:03.720
we should we should make a default so we don't have to do this comment you mean a magic comment
00:21:10.679
yeah this com this like okay how many people in here
00:21:16.360
do you put this at the top of all your files yeah does it suck you have to put
00:21:21.880
this yes
00:21:34.159
yeah but this this feature so this if I remember correctly this feature it doesn't break anything right now it just
00:21:39.799
warns you that you could be mutating it and it gives you the opportunity for you to put this comment but you put you can
00:21:46.880
specify false here so you if you don't want to change your code you can just say I I do not want Frozen string
00:21:54.080
literals
00:21:59.320
yeah if if you really if you really need you can also run the whole interpreter
00:22:04.600
with uh disable Frozen string literal so if if you have some Legacy application
00:22:10.240
and don't have the ability to go through and fix everything you should still even when this becomes default someday you
00:22:15.600
should be able to turn it off all right so to be honest I'm not a
00:22:21.679
fan of this change and I think uh Ruby is free so Ruby sing also free for
00:22:30.360
modification by default so I don't like this change I think uh M was wrong about
00:22:51.400
this what yeah so just now we have a warning
00:22:58.640
for the uh modification to the literal just now and there are some reports
00:23:05.360
there oh so many such a warning so so act at from actual application I'm not
00:23:12.480
sure but so with this report you don't change your mind just now not yet
00:23:22.240
be yeah mostly because of the so many people already use the the magic comment
00:23:28.039
true so the use the string Frozen string literals and the second things is that
00:23:33.919
we still expecting the performance Improvement by the uh the the Frozen
00:23:41.200
string lead oops uh actually the the when I got this idea first I expecting
00:23:49.440
the the much more uh performance Improvement because of the uh the the
00:23:56.200
lower GC pressure for string or something like that but in reality this
00:24:01.880
is not that much yet so that yeah I
00:24:08.240
hope I still hope that still we still have the room to improve the
00:24:29.840
that's Lake test task adding DW atically
00:24:35.240
oh does it h I think we shut it off in rails I'm pretty
00:24:40.360
sure don't we I don't think so you don't think so no like the default like I think the default test tasks don't run
00:24:46.520
with dasw it does it oh good only on c not
00:24:54.720
oh only on CI apparently but no that's true for rails
00:25:03.200
itself but rails apps I don't think it runs with d yeah I don't think that runs
00:25:08.559
with DW I think I think it does does it does anyone
00:25:15.799
know okay uh oh this is my secret branch
00:25:30.919
yeah don't get too excited please let's
00:25:37.720
see how do I make a new app bundle EXA
00:25:56.360
ties sure oh we got to do Dash Dev
00:26:07.039
okay oh this is going to take
00:26:13.440
forever let's go back to the this guy all right so what do what do we
00:26:18.960
think here Matts we're going to keep this yes
00:26:26.440
yes yeah yeah under current situation yes but okay
00:26:32.240
yeah I'm not fully satisfied though but yes uh all right our next question here
00:26:39.360
is from koichi it says to Shopify Ruby core team I guess that's the four of
00:26:46.039
us these three plus me um how how to
00:26:51.240
communicate with each other and how frequent do you share the details of each other's projects
00:26:57.240
um very should I answer that you guys care you guys want to comment you
00:27:05.919
comment is that on now okay uh we have like two three meetings a week sometimes
00:27:12.840
it's too many but we talk about that and then we have a little geekbot thing that
00:27:18.840
asks us about our project and then if we need more meetings we have more meetings so like I would say communicate a lot
00:27:26.399
about our projects
00:27:32.279
we have we have weekly weekly meetings between all teams do you make a a big
00:27:38.399
goal for for some period for example a one year goal or half year goal or
00:27:46.720
quarter goal or something like that our
00:27:54.000
process somewhat gets contaminated by The Wider project management process I Shopify so
00:28:03.279
uh we do the longer term planning through that I'm not I I can't go into the
00:28:09.840
details because it's it's the side by the higher ups they kind of come down do this sort
00:28:17.720
of planning and there like set dates and whatever but also at a lower at the team
00:28:23.080
level we have our own internal GitHub board where we prioritized that
00:28:32.000
way so I also want to add that uh each of us as individuals have a lot of
00:28:37.440
autonomy on what we do and for example um when when I have ideas for new
00:28:44.080
projects I do prototypes and MVPs for a couple of days maybe a week um to
00:28:50.279
demonstrate the feasibility and the benefits that we get and um and and once
00:28:57.279
we demonst rate that we can propose those as projects to to our leadership
00:29:02.360
to to our management um and they formally become projects depending on um
00:29:09.240
our capacity as like you know how many projects we have going on at the same time um and and based on priority um so
00:29:18.080
you know some examples of this is work that I've done in the past with uh memory management inside of Ruby such as
00:29:24.960
the Ruby M Che gem or the Ruby fre at exit feature that was introduced in uh
00:29:30.360
Ruby 3.3 to um make tools like valind or asan address sanitizers by Google uh
00:29:38.000
those tools to work on Ruby so you can find memory leaks inside of Ruby or in Native gems such as
00:29:44.519
noiri um and whatnot and uh those were features that that you know I had an
00:29:51.640
idea my mind and they weren't in any sort of formal uh planning but uh we we
00:29:58.159
worked on those uh because uh like people on the team like demonstrated
00:30:05.120
that these features uh are are useful and they can be implemented within a few
00:30:10.720
months see thank we're blessed with great
00:30:16.440
managers I see like goof so the background of this question
00:30:22.000
is that the Ruby is open source software and there are no managers
00:30:28.640
so maybe this is the first time to uh get such a sophisticated work row so I I
00:30:35.840
want to ask us how to do that so this this was background thank
00:30:41.519
you okay let's move on to the next question what is your development
00:30:46.799
environment coichi asks you are using sir
00:30:53.039
ma no I prefer Linux Linux or
00:30:58.799
what were you using lenux I'm MAC Osos but have access to those Cloud things
00:31:06.960
for Linux I I use a mix of Mac OS and Linux I have a Mac OS machine but it's
00:31:12.760
kind of like a my dumb terminal I'm usually that's aent to Linux machine but then if I'm on an airplane I'm
00:31:19.600
developing on Mac OS so both and and Vim
00:31:25.679
yeah Vim everyone use
00:31:30.840
beam no no what do you what do you
00:31:45.480
use I I use a mix of Mac OS and Linux
00:31:50.679
2 um my because work supplies me with
00:31:56.519
the laptop and I like the immediacy of the the responsiveness of typing straight into
00:32:04.080
the laptop I use maos but for some
00:32:09.639
Advanced tools like RR which allows for time traveling debugging those are only
00:32:15.039
available on metal machines on Linux so even with like if you rent a VM with AWS
00:32:22.840
for example that wouldn't even work so for for specific tasks like say you get
00:32:29.039
a bug report that's Linux specific you have to use Linux it's always
00:32:40.279
mix Maco is an envm nice
00:32:47.440
yes I think I am only one uh oh okay too
00:32:52.639
uh I I I use uh windows and uh but
00:32:58.519
uh Native Windows is not uh for me so I'm using vir machine called
00:33:06.039
WS2 and as I edor uh I I was long fan of V
00:33:13.480
but recently I use uh V code okay I also
00:33:18.919
use a wl2 on Windows and they're using the uh emo Chrome editor nobody use the
00:33:27.080
XY Z that's
00:33:39.360
Linux and then my desktop is Linux and I use
00:33:46.679
emx I use emx I use emx okay we have one emac user
00:33:54.080
oh me uh I use I use Mac OS and Linux but I use
00:34:01.000
um what what questions should we do here I feel like we should take some from the
00:34:06.440
audience should we do this one yumian let's do this one how do we make
00:34:14.000
strings faster in MRI any ideas
00:34:29.800
the one the one idea is the for the very big and long strings that we can use the
00:34:35.119
data structure named rope but uh I have no other idea
00:34:52.320
ticket the string adoption yeah string adoption uh uh how do I search this
00:34:59.960
thing yeah I didn't write I haven't write the answer yet to the that issue
00:35:05.359
but I don't like the name adopt
00:35:28.800
where is it a do I don't see his ticket lat issues latest issues
00:35:41.280
okay oh
00:35:55.119
okay no
00:36:00.400
no uh adopt
00:36:06.560
okay thank you
00:36:11.920
20878 there we go okay so I'll I'll describe this for
00:36:18.440
everybody here in the audience what this is is a c API that creates a string
00:36:24.240
object uh given so you you give it a uh list of characters in C and it
00:36:31.520
creates a new string object with that the difference between this API and our current API cap API is that when you say
00:36:37.760
hey make a new string for me please and C it'll take that string it'll mem copy it into its a buffer that it owns in
00:36:44.200
this case we're saying Hey I want you to make a string object but I want you to take ownership I want you to take ownership of these bytes so the idea
00:36:51.319
behind this is that like say we read something off of the network uh every time we want to make a new Ruby string
00:36:56.760
we have to copy all the btes that we read off the network in this case we're saying hey I don't want you to copy those bytes I just want you to own them
00:37:03.119
now so it allows us to create new Ruby strings without doing a bunch of mem copies so that's one
00:37:10.720
idea but apparently Matts doesn't like the name adopt so but you're not against the
00:37:18.240
feature in general yeah
00:37:24.839
okay all right let's pick another another
00:37:30.040
issue uh sure yeah let's do okay yesterday you
00:37:38.160
touched on autot typing being introduced to Ruby what would that look like and what does it mean for
00:37:44.880
RBS uh the r RBS is the part of the technology that forms the the automatic
00:37:50.880
typing probably okay so it'll help with autot typing
00:37:57.160
okay
00:38:03.680
uh any plans for better rust extension support EG official bindings and the
00:38:11.079
like I think the answer is
00:38:16.880
no uh do Alan do you want to comment I have strong opinions about
00:38:22.800
this because in the company we have a bunch of people writing Russ extensions
00:38:28.599
and have Labor under the the false assumption that they're safe and usually
00:38:37.680
when it it's a big red flag because they we we've ship rust extensions to
00:38:44.680
production and they've fallen over multiple times and it's in and the
00:38:51.000
categories of bugs in there are the type of bugs that you would think rust solves
00:39:00.040
so I'm I'm biased against more rust extensions and the the worst part is the
00:39:07.599
the people riding the the Russ extensions don't have the rec of
00:39:14.000
knowledge debug those problems and the production debugging ends up falling
00:39:26.599
so I would pref for uh at least at least not in the
00:39:33.160
current shape of the the rest extension landscape that we add official support I
00:39:39.079
think if we want to a lot more work needs to be done in that
00:39:47.119
area oh yeah sorry said
00:39:58.839
so Ruby right now Ruby's API is a c
00:40:04.880
API and and C++ too but mainly C and the C++ part was added less recently too so
00:40:14.720
as soon as you switch into rust you've given
00:40:20.560
away your the ability to get first party support from Ruby Corp because from our
00:40:26.920
perspective Ive we're supporting the Capi
00:40:32.160
and you know so you're sort of a tear down anymore like whatever support you
00:40:38.520
get is filtered through a third party whoever is maintaining the infrastructure that's doing the C to
00:40:45.040
rust translation and in that
00:40:51.119
process bugs are introduced um because because the
00:40:57.760
the capis they have contracts sometimes they don't get transferred over to the r side correctly
00:41:05.400
but the contracts are contracts if you break them unknowingly your whole app
00:41:10.960
will fall over so that I think that's part of the
00:41:16.800
reason why the rest extensions haven't been of great quality of the the ones
00:41:22.240
I've seen
00:41:27.480
um they fall over due to like cod process crashes memory
00:41:33.079
corruption um concurrency issues like uh there's there's like a very sharp API
00:41:40.440
where you could release the global interpreter lock and if they don't use that properly you get heart to debug
00:41:46.319
action at a distance bugs that sort of things so one of the things that Russ
00:41:53.520
guarantees is that uh it frees all of your memory for you using the borrow Checker so it knows when things go out
00:41:59.760
of context it knows so then that memory can be released back to the system and one of the things that uh it's very hard
00:42:07.640
to um think about in Ruby are uh exceptions so you use exceptions in Ruby
00:42:14.640
you raise it and and you know everything seems easy but in the Capi that's
00:42:20.160
actually it it it's it's actually very hard to reason about how exceptions work because how exceptions work under the
00:42:27.559
hood in Ruby is that when an exception happens it does what's called a long jump to the um next Ruby frame to the to
00:42:36.640
the bottommost Ruby frame and uh that means that it skips a whole lot of um of
00:42:43.319
native frames that you have either in C or in rust or in C++ or whatever and that circumvents all of the
00:42:50.760
borrow Checker stuff so everything that you've allocated will not get
00:42:56.160
deallocated and because it will immediately jump to the last frame the
00:43:01.200
last uh C frame in that corresponds to a ruby frame inside of Ruby and so if you
00:43:09.119
use anything inside of Ruby that could possibly raise an exception uh or or raise an exception yourself um then uh
00:43:17.359
you have to make sure that everything that that the borrow Checker had that
00:43:23.079
that that that uh every memory that you've allocated within all of those frame games will have to be free or
00:43:28.880
otherwise it will just leak and so those are guarantees that you think you have
00:43:35.000
because you're using rust but because you are using that as a c as a ruby extension you actually do not have any
00:43:41.720
of those guarantees yeah it's also true for C++
00:43:46.760
and it's also a problem for C programs because it could still long jump and you could you could leak memory so it's a
00:43:53.119
problem for everything but uh at least in see your you're a lot less confident
00:43:58.520
because you don't have these guarantees but in Rust you feel like you have these guarantees but you really don't yeah I
00:44:03.880
think I think the problem is with rust extensions it's the like you have a false sense of security that rust is
00:44:09.559
going to deal with all this stuff for you but you actually have the same foot guns and gachas in Rust when dealing
00:44:22.800
so okay let's do another anyway I don't know so to answer this
00:44:31.680
question do we have plans no I don't think there's any plans Matts are there
00:44:36.920
plans for better rust extension support uh yeah the few uh few of us use Ras as
00:44:45.880
a primary language so that we knew little about the Russ uh in
00:44:51.319
interoperation between C and C and and rust so that means
00:44:57.880
the we we cannot provide the better r support from from us so that if someone
00:45:05.280
contribute the ideas or the implementation for the r integration to the Ruby so that that I'm not against
00:45:13.200
about integrating them but uh I I don't think we can provide by ourselves so
00:45:20.000
that the better rust uh extension support the in my opinion so that uh if
00:45:29.359
we can uh to provide the r feature to Ruby you mean that if rust extension
00:45:36.319
call back to the the ru VM does not call back to the ru VM there's easy way to
00:45:42.680
integrate rust extension but if the the rust extension tries to the interact to
00:45:50.960
the RM back and force it is far more difficult due to the the issues just
00:45:58.559
just Alan then
00:46:05.280
explained all right yeah and exception of
00:46:11.319
course oh ropes simd support ah
00:46:17.319
oo oh this is a good idea yes simd
00:46:22.760
support uh so simd simd stands for single instruction
00:46:28.760
multiple data yeah right so it's basically like one instruction that lets you do things it's a type of parallel
00:46:35.040
Computing that you can do so you can do stuff faster basically so I guess we
00:46:41.000
might use this for code ranges yeah analyzing
00:46:47.640
strings um let's pick what should we pick
00:46:55.680
oh oh wait what is this what is this one did we do this
00:47:00.760
one how about this feature sure let's read it 2087 I have it
00:47:10.720
memorized Atomic initialization for rapor local storage I don't know this
00:47:16.559
ticket yeah it's all up to you so I I asked uh so in advance I
00:47:23.559
asked about this feature yeah in the last you don't like this name maybe or
00:47:31.359
is it acceptable you rename the local variable
00:47:36.440
to Stage right local storage in it yeah yeah yeah local it much better
00:47:46.280
acceptable yeah and yeah it's okay but uh but uh but the
00:47:55.480
the term in it has the you know several meaning could have several meaning like
00:48:02.240
initializing or your assignments or something like
00:48:07.599
that yeah the does in it only is
00:48:15.200
oh or
00:48:21.359
something so this this is how to implement so maybe usage is so
00:48:27.440
doize this one what is this
00:48:34.480
for what explain the background so this method uh so we we have a left local
00:48:41.680
storage and we can get and uh lead the uh from the uh local stage already we we
00:48:50.319
have already feature but we do we don't have a uh way to set the the initialize
00:48:58.480
atomically so for example uh on RoR we can run the multiple threads and so the
00:49:06.400
many St can uh can initialize one local
00:49:11.480
uh stage so it is a problem and there is no way to prevent such a situation so
00:49:19.240
this method is uh provide for this uh to
00:49:24.480
initialize uh the some data in atomic in
00:49:29.760
uh in atomic car so for example so we don't need to
00:49:35.079
this feature for the uh the without rure
00:49:40.520
because we we we can initial at the requireed
00:49:45.680
timing but if we want to uh initialize laily so after after
00:49:54.319
require so the any any SL can initialize this data uh so this kind
00:50:01.760
of data so we need to make a way to initialize only once so this is same as
00:50:10.160
a regular Expressions once
00:50:18.599
option every time in the Inc stands for
00:50:24.520
increment method Cod yes the it tries to call the local
00:50:31.760
storage in it m and uh what once if we
00:50:37.960
don't have the local storage named MTX MTX then in uh
00:50:45.559
initialize with block but if we if we already have the local uh storage named
00:50:52.880
MTX the the block is simply ignored you the MTX
00:51:03.280
data right I I can understand your intention and then then the local Str
00:51:17.200
named then Co synchronize to the return value
00:51:37.680
so you have two local storage named MTX and the count yes but uh uh you only
00:51:47.480
call the local storage in it to the MTX because it should
00:51:54.520
be I don't know atomic or something so the count also initialized
00:52:01.319
at the beginning in this
00:52:13.160
case yes so that's the begin begin so
00:52:18.200
the larg begin you may you Capital begin and there uh maybe so so it shows that
00:52:25.040
there uh uh uh every RoR run this uh begin block for
00:52:36.240
example for yeah uh if it is needed so begin uh forc
00:52:45.200
to run every uh RoR but in this
00:52:50.240
case well we can we can initialize very easy
00:52:56.839
so to for me to understand now I understand the
00:53:04.079
intention but uh I don't think the name the provide enough information for
00:53:11.920
the behavior because the the you only provide the name in it stands for
00:53:17.240
initialize but it has the some kind of the the conditional check if the uh the
00:53:24.200
the storage name is already initial ized or not that would be that would be you
00:53:32.240
know not intuitive name for the behavior I think because of the you know the inet
00:53:41.799
usually stands for initialize but uh this one you know initialize that but uh
00:53:49.839
in addition if the the the strage is already initialized it just ignore them
00:53:57.640
so that that that conditional is not included in the name that is kind of
00:54:03.160
problem I think do you have a good name of I'm not sure yet I see I see all
00:54:08.839
right so the P so the it is not accepted yet I understand the behavior and the
00:54:15.640
requirements but I don't agree with the name all right so the naming issue naming issue okay and hi
00:54:33.960
exactly yeah probably he wants to be a some kind of the lazy initialization so that if we don't call
00:54:41.760
the if we don't call the say the ink method the ink method at all so that we don't
00:54:49.440
initialize the MTX at all so that that's what he that's what he needed I and once
00:54:57.599
is another terminology I agree that in this case so anyway so concl we have
00:55:05.640
conclusion about that so name is not okay is anyone putting this in in hackmd
00:55:10.839
I'm not it's thank you it's it's it's kind of like a div was meeting yeah yeah so
00:55:17.640
okay uh we held a developers meeting uh in the month one once a month so and we
00:55:27.480
discuss each features and there so mosa and other developers discussing
00:55:34.720
features and names and so like like this time so there are many name is not
00:55:44.319
satisfied I don't like the name okay anyway I like this question
00:55:50.920
other than Ruby what is your what is your favorite language
00:56:01.480
I can't uh pick one list
00:56:07.559
them I know many yeah I have uh I have
00:56:15.119
wrot I have written qu in many languages
00:56:20.480
and 128 langages but so I can't pick one
00:56:31.760
I'll skip this question x86
00:56:39.119
64 ideally with avx2 AVX 512 support
00:56:44.480
anyone no uh controversial but I actually kind
00:56:53.200
of really enjoy Java there's a lot of there's a lot of very nice things a lot of very good Tooling in Java but it
00:57:00.760
works the the the typing system makes sense and what is well thought out looks
00:57:09.160
elegant sorb uh and uh and and yeah but you know there's
00:57:16.760
a lot of a lot of downside of java but so does any other
00:57:24.119
language oh I don't I don't no I don't write a lot of other
00:57:30.119
languages like see I was going to joke that it's like
00:57:35.799
uh whatever you write make files in since I spent so much time doing that I'm like now a make File Expert it's not
00:57:47.480
in uh I earnestly see uh I really like it uh but I can't defend that choice
00:58:00.119
I I also like C because it is easy to understand what there become to the
00:58:09.319
machage Matt as a programmer I like c as a
00:58:14.839
language designer I like L
00:58:20.280
mod me I like me I like Zig yeah
00:58:26.839
because I'm a hipster why do you like huh why do you like Zig why do I like Zig uh it's
00:58:34.119
it's nice like it's easy it's easy to write and it's less like the I feel like
00:58:40.039
if I try to use it's like using rust but the compiler doesn't complain all the
00:58:45.520
time yeah much simpler and then the more closer
00:58:51.039
integration to c c yeah yeah it's what I really really like about it is I can use so I can use zig and call into C
00:58:57.039
libraries but I don't have to use like bind gen or anything it just it just works see H oh it even catches cbgs yeah
00:59:06.240
I so I compiled I compiled prism with zigg and then zigg caught a bug in prism
00:59:13.240
so I was able to fix I was able to fix that so zig's compiler is C compiler is really
00:59:19.240
cool I can easily Crose compile
00:59:24.640
everything okay that's all of us uh let's do next question Matt yes uh you
00:59:34.119
misunderstood the behavior of Colonel sleep in the game show how about changing the
00:59:40.359
Behavior Uh yeah
00:59:46.799
uh the the behavior is the the the
00:59:53.319
in sleep Returns the the seconds
00:59:59.359
slipped as a return value the the point is I thought the integer to float
01:00:06.760
conversion is the normal casting like a cut off but actually it rounded okay
01:00:13.119
okay okay yeah that's my misunderstanding uh actually the
01:00:18.920
original version of the sleeve the which I wrote was the uh the used the plain
01:00:26.000
sleep Library fun C library function which returns a second so that the unit
01:00:31.079
is the integer then later the someone I
01:00:36.240
don't remember who but someone extended to the the to accept the float value as
01:00:42.960
a you know the more precise time and then at
01:00:48.680
then we kept the the return value as a integer as a some kind of the contract
01:00:55.400
signature so then in that that's what I remember but I I don't remember how to
01:01:02.280
convert the flo to integer like I thought it is cast cast the cut off and
01:01:15.760
arounded okay so are we going to change the behavior no uh yeah momy is talking
01:01:22.760
about that all right Let's see we have more
01:01:27.920
questions at the bottom but I'm going to be really selfish right now and do this one uh because this is my my
01:01:35.720
question um so I filed this feature to add an environment variable for changing
01:01:42.640
the thread Quantum this is this is the ticket so
01:01:50.079
basically um I'll do a really quick I'll do a very very short presentation uh who here so who here
01:01:57.200
went to Evo's talk or to koichi's talk okay a few people all right yeah
01:02:04.079
you went to your own talk great so we have a we have an issue like
01:02:09.359
if you have two threads like T1 um and T2 and they're running at the
01:02:15.039
same time but we don't actually run at the same time because we have a gvl so we can only get CPU e either one of
01:02:21.200
these threads can only control the CPU one at a time but we have a certain thing where you can do like IO so let's
01:02:27.760
say T1 is doing IO for amount an amount of time uh while it's doing IO we allow
01:02:34.359
T2 to execute so T2 is allowed to do CPU time here CPU so it can run but we have
01:02:41.039
an issue where uh when we switch between threads we switch on a Quantum which is 100 milliseconds so let's say T1 is
01:02:47.680
doing CPU time it's allowed to do that CPU time for 100 milliseconds before we switch to T2 now the issue is if we have
01:02:55.559
two things running truly in parallel here like IO and CPU let's say this CPU time here starts right here and it's 100
01:03:06.039
milliseconds for our Quantum now this IO time we're not allowed to switch back to T1 until this
01:03:13.079
Quantum finishes here so the amount of time we get to come back return to T1 is
01:03:18.680
actually um IO time plus
01:03:24.240
Quantum yeah so
01:03:29.599
my problem is that I think 100 milliseconds is too long so I want to shorten that and I
01:03:37.079
want to do it with a environment variable and that is what this is about yeah
01:03:43.200
okay so I we can merge this so so uh we
01:03:48.760
have uh the current 100 millisecond time Quantum is not has no strong opinion uh
01:03:58.440
so there is no uh strong reason to pick to determine this uh number so I I we no
01:04:07.559
maybe nobody again to introduce this environment VAR I think okay so anyway
01:04:14.119
anyway anyway anyway you you you uh your sample program is not doesn't make sense
01:04:21.880
to show their the problem so yes I want yeah Evo so Evo made a
01:04:28.720
comment showing the showing the actual problem so if you see yeah I
01:04:33.960
understand uh uh problem but uh you you
01:04:39.760
it doesn't make sense sorry I emphasize jeez
01:04:48.440
okay and also we need to discuss more about the name of enable oh my gosh okay
01:04:56.359
the name of the environment all right maybe we can we can uh accept that all right okay uh I'm not strongly against
01:05:06.680
uh this proposal but I'm a bit afraid that uh uh of the overhead uh because Too
01:05:16.440
Short uh St cor May because uh overhead
01:05:21.520
so did you uh perform uh performance evaluation uh we haven't we haven't
01:05:28.319
tried performance evaluation on this yet I think it is needed first I
01:05:38.680
think so that yes the question is could it be available but 100 by default that's what this patch is is it doesn't
01:05:45.000
change the default so it's always the
01:05:51.520
same that was the idea is we default to 100 and then and people can tweak it
01:05:56.960
depending on what their application does because the problem is the it depends on what your app is
01:06:03.440
actually
01:06:09.559
doing okay all right so yes we can merge this
01:06:17.279
but with a different name yeah we usually use the say
01:06:23.400
something limit or as but as a name of the parameter like
01:06:29.680
environment variable but you know I'm I'm not sure the the quantum is problem
01:06:35.359
not okay yeah 100 millisecond is the the the
01:06:41.799
ancient the Linux contact switch timing I don't I think that it isn't anymore
01:06:47.400
yeah it's it's it's no more it's a yeah I think it's dynamic or or something
01:06:52.440
yeah I understand I think theuer also changed with within the last year again so I have no idea yeah I have no idea
01:06:59.559
too but it you know it's originally taken from the theux process Contex
01:07:05.200
switching timing okay all right I'll propose a few other names and then add
01:07:10.880
it to the dev meeting thank you all right um this is a good question
01:07:16.520
how everyone how did you end up building programming
01:07:21.799
languages I think Matt should start with this one how did you how did you start building programming languages uh I want
01:07:28.240
just wanted you know you know for most of the
01:07:33.520
people that language uh programming language is given so that you learn the existing programming language and write
01:07:40.160
programs like games applications and something but uh for me I I want to
01:07:45.799
design the interaction between computer and human this is so that to to
01:07:54.039
accomplish that purpose of the design designing interaction so that I I like to design my own programming language
01:08:01.039
that's the whole purpose and then then uh why I wanted that
01:08:09.720
that's for
01:08:15.719
Joy no I think you should go kigi yeah uh when I was a uh University
01:08:24.600
student I went to academic uh uh academic
01:08:30.199
lecture course about garbage collection yeah and I sit down at the table and the
01:08:48.480
there that's all that's all yeah I recruited
01:08:56.799
the reason uh why I I was uh involved in
01:09:02.000
uh rby development uh is because Ruby was buggy
01:09:07.520
so especially when I uh was I became
01:09:13.040
a y y interpr was introduced to Ruby master
01:09:20.640
branch and it was very buggy so I was enjoyed
01:09:26.000
I I was uh enjoying the Kings and uh I became a
01:09:34.319
commit well I was just say reals developer but luckily I I was born in
01:09:41.120
Japan I lived in Japan so I could see and directly talk to the language
01:09:48.120
creators I love that experience and I'm still in love with that experience I
01:09:53.199
can't quit and that's what I'm
01:09:58.320
doing uh not really a language builder um I got my start building a compiler
01:10:05.520
but I didn't really design the language it's more a clone
01:10:11.280
of things I already thought I was mainly interested in the the
01:10:18.480
mechanism inside of the compiler more than coming up with language constructs
01:10:24.840
that people my own M like I don't particularly have many opinions in that
01:10:31.280
area and um once I got started at
01:10:37.440
work we had a bunch of Ruby crashes on Mac OS and I sort of felt this
01:10:45.040
indignation that why would it be crashing like it's so simple just go fix
01:10:50.640
the bug and that's sort of how I started
01:10:57.640
I'm also not really a language builder I mean I just make the language faster but
01:11:03.199
uh I was originally just a a rails Dev um and then four and a half years ago I
01:11:08.920
joined this team at Shopify where they were like make Ruby faster with this
01:11:15.800
idea and I was just a little intern and I was like what uh and um and uh you
01:11:25.000
know for the fear of not losing my job as an intern um I I learned how to do
01:11:40.920
am yeah the person who assigned my work is sitting right
01:11:46.679
here uh I would say in my case it's Aaron's fault um cuz he I found a bug in
01:11:54.480
reals and he he was like why don't they fix that so we ended up pairing on it and
01:12:00.639
then uh and then he pushed all of the stuff he didn't want to work on onto my
01:12:05.840
plate and here we are um similar for me I think I I wrote
01:12:13.760
Ruby for 10 years before getting involved in improving the language and I think I just didn't realize that it was
01:12:22.040
possible until then so which is also aon's fault thank you uh did thank really all right
01:12:31.360
I feel like I've cursed you more than um I'm I also don't consider myself
01:12:37.040
to be really a a like a programming language builder I just was really interested in how Ruby itself worked
01:12:44.000
dove into the code and like tried to improve it I mean I just like basically I learned the language I
01:12:50.639
fell in love with being a Ruby programmer I learned rails I fell I really enjoyed building websites and
01:12:56.280
stuff with ra building applications with rails but I wanted to know like I'm just very curious and I want to keep like
01:13:02.159
going deeper into how does this thing work and then like okay well now I understand that well how does like how
01:13:08.040
does the thing below that work and then I just keep going deeper until I
01:13:13.080
understand as much as I can so I did say in my talk that learning sucks but that's not true I'd actually like
01:13:21.040
learn uh let's see uh everyone how much time do you spend building non-trivial code with languages besides Ruby and C
01:13:28.960
or rust for Ruby Dev and how does that affect your language design and direction of Ruby oh my
01:13:34.600
goodness wow mattx do you build any applications with languages besides Ruby or
01:13:43.120
C uh some years ago
01:13:50.040
I I wanted to switch to the G Gmail when Gmail introduced in 20 four or something
01:13:56.560
but uh at that time I we we didn't have the some kind the universal internet
01:14:03.239
access so that uh during on the plane or maybe in the in in the in the mo you
01:14:11.239
know in the transit so that I we didn't have the the internet access at the time
01:14:17.600
at the moment so that I decided to write so my
01:14:23.000
own mail reader to work just like a uh the Gmail like a
01:14:29.920
you know no no folders no uh no classification just a full Tex search
01:14:37.440
then things then I wrote the back and the in Ruby but uh it's I wrote its
01:14:44.840
front end the interface in emox
01:14:50.120
list and I still use it so that oh the old the all my email is uh handled by
01:15:09.840
as a main reader but uh when I write replies I use
01:15:23.639
build anything with besides do you use any languages besides Ruby or C so the uh so recently I made a
01:15:32.360
deug so deak of Ruby in Ruby so I feel that so Ruby is very fantastic language
01:15:42.000
fantastic language so very easy to write yeah yeah I agree and we can find the many many
01:15:49.920
things we need to uh Implement for to uh in in C
01:15:55.480
to support
01:16:05.040
de not Ruby not hor Ruby yeah uh uh in my hobby time uh I uh enjoy uh
01:16:16.560
development uh web front end so I uh enjoy typescript and JavaScript but uh
01:16:24.480
to be honest I'm not sure it how it affects the design decision of
01:16:43.360
game mind sweeper I don't know where it is this
01:17:03.840
I think yes that this is a Min swe G but uh it is uh
01:17:12.440
automatically uh selected by AI so uh please uh pick any s yeah and uh yes
01:17:26.840
so you you can so just uh like testing
01:17:34.639
so brain testing part is automatically done by Ai and you can just enjoy luck
01:17:43.880
testing Pi picking random s by your
01:17:52.120
Ro so this is one of the my uh joke
01:18:07.840
project I don't know I'm trying to do everything with
01:18:14.719
Ruby uh similar to the emac list situation I use neim which uh comes with
01:18:24.080
lu script um and also it interacts with wind
01:18:30.239
script I do some of that um and also with Lua I use this window
01:18:36.600
management Window Manager for uh xor called awesome
01:18:43.560
WM it's scriptable in Lua and you could do whatever you want with the windows on
01:18:50.199
screen like say you want to have some presets or whatever the way windows lay down you deal that DOA so um and then
01:18:59.639
when it comes to how how it how it impacts my language design skills not
01:19:06.920
very much if anything it makes me appreciate the the myriads of batteries
01:19:13.280
that come with Ruby uh I Lua is very
01:19:19.320
it's a very simple scripting language in in the sense that there aren't that many moving parts so you don't get the
01:19:26.280
batteries filed in I often find myself wanting something that would be in Ruby
01:19:32.120
but need to look around to figure out how to do the same thing in bua but um I
01:19:39.639
appreciate the Simplicity of the language similar to
01:19:47.199
how every mostly everything in Ruby could be explained by Method calls
01:19:55.880
I like that part of FL so there is this one game that we
01:20:03.080
allowed to expense at work it's c factorio and a couple of weeks ago I
01:20:08.560
decided that I was bored and I was going to get it and let's just say it has become my second full-time
01:20:22.000
pay me and and uh so it how it affects my design
01:20:30.360
and direction of Ruby well definitely in a negative
01:20:37.360
way uh I don't remember the last time I worked on something that wasn't C Ruby
01:20:44.880
or RS uh one time I wrote a apple shortcut
01:20:50.280
to scrape pelaton site to put it in a spreadsheet so I knew what I workout so I was going
01:20:55.960
to do that week but I haven't used it in a while and that required me to write some JavaScript which I did not enjoy
01:21:04.760
so yeah similar stuck with try to do as
01:21:11.840
much as Ruby as possible oh I'll go okay uh let's see
01:21:18.400
I'm been trying to write stuff in R like whenever I need to do graph graphs I'll use R um I wrote a I'm very proud of
01:21:41.120
something this it's a Vim plugin that converts
01:21:48.280
your code to RTF so basically like you highlight the code and then you
01:21:55.800
do two RTF on it and then you can paste the code into keynote and it is syntax
01:22:04.199
highlighted for you so and you can like highlight stuff
01:22:10.199
and then copy it as RTF and then um paste that into keynote and it is all it
01:22:16.760
just works so this is my very first non-trivial Vim Vim script plugin and
01:22:23.199
I'll tell you working with Vim script makes me realize how bad of a language it is it's really bad and I love Ruby so
01:22:31.120
much so yes
01:22:36.760
yes oh with neovim it comes with scripting support for including Ruby too
01:22:43.560
so you could put all your wi script to Ruby if you really want V Vim you can
01:22:49.159
you can also do Ruby from Vim but it's like the it's just bad it's not good
01:22:55.440
all right uh next question everyone what's the largest in terms of memory usage Ruby application you've ever seen
01:23:01.880
does anyone run Ruby with with 10 gigabyte plus heaps
01:23:07.159
successfully what Peter uh I mean it's probably shop by
01:23:14.199
core maybe not I don't know our monolith I don't know is it it's not even that big how many G how many gigabytes do we
01:23:21.719
use do you know I think only like two or three gigs of memory like I don't think we use that
01:23:29.040
much the the our test Suite uses in that ballp part like seven gigs
01:23:36.320
10 gigs and sometimes it runs out so it's like it's a pretty tight bound um
01:23:44.040
and also Peter you must be familiar with what happens when you run Ruby with
01:23:50.159
asan balloons the memory um so yeah yeah if that
01:24:02.880
counts when maybe 10 years ago 10 more more years ago so the big data was Buzz
01:24:13.360
was so I bought uh
01:24:18.800
20050 uh about two 250 gigabyte machine to try the Ruby
01:24:27.960
with the hug memory consumption so making the many dummy data on the such a
01:24:36.080
machine so and we I found find out that the the termination of the Ruby
01:24:43.840
interpreter is very very slow on such a case so maybe there are many many uh
01:24:50.480
inefficient uh uh algorithm in that will be interpret with big data
01:24:58.239
so so I so I don't have actual application but maybe 100 gab Ruby
01:25:06.560
application has many issues I think it
01:25:15.360
comment well there has to be a production like fluent z uh
01:25:22.400
server production I guess I haven't seen myself but
01:25:34.520
yes uh at work we run a a simly large like 1.5 gig maybe at at the largest
01:25:41.239
size so not not dis similar at work is at GitHub um so so similar to Shopify
01:25:47.040
model it sounds like a little smaller um sometimes I I have a tool that analyzes
01:25:53.159
the Heap dumps from that so since it has to have several objects for every object the main application
01:25:59.360
has in memory that definitely gets to 10 gigs plus all
01:26:06.719
right uh Aaron how close is your rapor friendly rails Branch not at all
01:26:14.800
um basically kichu kichu is saying like oh we can't we can't run rails with
01:26:21.800
rators and I want to know like why specifically because I don't know the reason why so I started like trying to
01:26:29.719
run rails with reactors to see what we have to change to fix to fix it either change rails or change rators somehow to
01:26:37.119
try and investigate what the deal is um but it's not close at
01:26:44.080
all yeah eager booting so I was trying to eager boot eager boot the rails application make the like do make
01:26:51.520
sharable on The Rails app itself and then see what crash or like what blows up and then try to figure out like why
01:26:58.280
it seems like we have a lot of lambdas or something that can't be made sharable but I'm not sure why we can't fix that
01:27:06.639
so I'm investigating yeah clo like I'm not sure
01:27:14.119
why we can't use closures maybe kichi can tell
01:27:20.800
us the closure is modifiable yeah the exteral Contex can well it
01:27:29.040
references like it references it has references to external data yeah that
01:27:34.719
can be to be to be sharable the the object
01:27:40.440
should be recursively unmodifiable what can you modify in a
01:27:46.440
Lambda like I don't get it
01:27:52.320
Alan we had to take so I think right now the makes sharable operation on Lambda
01:27:58.679
is sort of a syntactic operation a part of it is in that if it
01:28:04.400
it's if it sees you writing a local variable to anything external to the Lambda it's whel rejected even though
01:28:12.880
that might be saved so we actually had a bug report recently where if they just change it so that the the access is to a
01:28:20.920
constant instead of a constant local variable and that it's not changed that
01:28:26.000
their program would have worked but the airor message there was somewhat confusing and they assumed that uh what
01:28:33.000
they wanted to do was impossible when all they had to do was change the
01:28:38.119
syntactic like jigle their what they wrote basically what it can you give me
01:28:43.840
an example what that is like a local variable write uh read even read even a read yeah I
01:28:50.880
think like this yeah so uh do it and then pass the uh the the lamb that to
01:29:03.119
exception like this yeah I'll fix it I
01:29:25.520
X to be a constant it should
01:29:39.080
not let's make this smaller
01:29:44.719
oh yeah it complains about the cell uh self right right right okay so
01:29:50.639
if we did like uh could do the stabby Lambda
01:30:15.560
situation yeah yeah
01:30:21.440
no why
01:30:28.159
why uh during the block the self is the OU ah the outer self okay okay okay uh
01:30:36.480
okay
01:30:41.760
um let's see right right
01:31:02.280
there we go there we go but if we do like if we do like
01:31:27.880
why does that work the block is off the block is Orphan so if I do can I do like xals 1 2
01:31:43.800
okay yeah so I think maybe what's missing is uh more documentation or some
01:31:51.880
sort of guide yeah it's pretty new yeah so it's because in this case
01:31:59.719
it's because we're mutating the outer we can mutate the outer environment right got it yeah the the problem you ran into
01:32:07.400
while typing out the program is also a usability issue right like accidentally
01:32:13.320
capturing self not sure about yeah yeah okay all right
01:32:27.440
I'm suffering through my my test okay did we hit all the questions I think we
01:32:32.840
did anyone have any more questions in the audience should we wrap this
01:32:39.520
up anything nothing okay oh wait no we got
01:32:47.000
some
01:32:52.239
you yes I happy to help you out of course yes another question over here
01:33:05.480
yes wait what is the
01:33:12.920
question ah do any of you use llms question was has anybody reap any
01:33:18.840
benefit from using LMS like chat UT or CLA or
01:33:26.560
copy yes I'll go first like it's
01:33:32.920
so it I've used chat GPT for like I used
01:33:38.920
chat GPT when I started when I was getting started with Zig and it helped out a lot especially because I as a very
01:33:46.360
very very much beginner but then the problem was like as I got a little bit
01:33:52.040
better uh the helpfulness completely dropped off and it was like wrong most
01:33:57.080
of the time so as my as my questions got more sophisticated its answers got
01:34:04.960
worse um I don't really like how AI mansplains code to me like I can get
01:34:11.679
that elsewhere so but I I do sometimes use it to rewrite text that I've written
01:34:19.320
uh fuga is going to find out that I fed my performance review into it and told
01:34:24.360
it to help me talk about leadership skills and then I rewrote it so it didn't sound insane uh
01:34:35.000
basically that's fine it went well I mean I didn't it didn't write it
01:34:40.320
for me I wrote it and it goite better and then like maybe I need it right now
01:34:46.239
to help me answer this question it's a lot of pressure having your manager in the audience
01:34:58.880
yes all right if there's no more questions let's wrap it up thank you all so much we'll come around to help you
