00:00:09.720
hey everyone how are you it's a early morning for me or for you
00:00:16.420
but late night for me if I drop dad's asleep here on a stage don't bother me I
00:00:22.000
will just sleep just carry me somewhere and I will sleep there okay today I want
00:00:27.369
to talk about sustainable architecture it's a really biggest topic what is
00:00:33.130
sustainable and what is architecture I will start with the introduction about Who am I and that question that everyone
00:00:40.750
probably asks themselves often I started
00:00:46.000
to work in raised rails later than akhira unfortunately for me I started with rails one something and I drew in
00:00:54.699
three years I accumulated lots of books a little bit of experience so don't judge me for that I like to speak on a
00:01:01.690
stage and I like to speak with people I really like hearing problems that people can face in their everyday life or is
00:01:08.800
there everyday software engineering so you have some problem that you're struggling right now find me I would like to talk with you and maybe help
00:01:15.790
maybe make your problem problem in move even worse so not and I can't promise you anything right now should be fine I
00:01:24.610
was working on various companies before so if you want to embrace the winter and
00:01:30.130
really feel what snow looks like and feels talk to me lots of snow really a
00:01:37.659
lot of snow we have a special road for snow today I won't talk today we won't
00:01:47.289
talk about what's what's driving application farther how we can grow our
00:01:52.840
application which time and what is them what is the main problems that we may encounter during the application
00:01:58.960
development this is talking fortune not about how you can do the perfect plane
00:02:04.479
even though I would like to talk about a perfect planes this talk not even about
00:02:09.789
their flying chair you can put a fan to your share and pretend it's fine but this talk not about it and it's not
00:02:17.950
about micro services and not about distributed monolith
00:02:23.070
however it's all about that and things it between about it and I will explain
00:02:30.190
why usually when you when someone have an idea an idea is to build something and
00:02:37.990
to make a project you will feel from outsiders as idea came to success we
00:02:44.080
know only successful stories but we rarely know about stories that end up badly unlike a Oh get down however were
00:02:51.880
lots of us work on a project that not not in a production anymore like at least me I work and feel them
00:02:58.770
because the ideas and success usually looks like this you have lots of
00:03:04.060
problems on anyway and you're trying to take this problems as you undertake undertake those problems and you face
00:03:09.280
the problems you're trying to build the solution those solutions they are not general they're usually custom because
00:03:14.590
your problem is constant because they feel that you you do in and you do right now for your project is something new to
00:03:23.350
the market it's something new to the world and those your problems are also new in your solution to those problems
00:03:29.020
also here and I rarely hear a phrase from people that we're running this new
00:03:35.709
project awesome idea and we're running it right now but only for three weeks and after three weeks we shutting down
00:03:41.500
the application we don't want to work anymore people usually think and think in a future they want to do the
00:03:47.980
application that will last for years that will help and solve certain problems for people and as you solve the
00:03:53.950
problems and you accumulate more features and you accumulate more solutions to your problem you grow
00:04:00.430
complexity and as you grow complexity you also accumulate lots of box and lots
00:04:07.570
of and lots of solutions that are custom for you as if star wars is only starting
00:04:12.670
the application you're thinking about well it's a greenfield project I can use whatever I like have you used the edge
00:04:18.299
edge case technologies and I will make it definitely the word the best thing at
00:04:24.490
a time at my time and then you finishing the project like in Midway of your
00:04:29.590
project and you seen that looks like I have a lot of things that I made incorrectly and I need to fix them but
00:04:36.400
just but rarely you get in back and you fix in the problems you're still building your perfect plane and you see
00:04:43.419
building things on top of the existing existing application and easy build that
00:04:49.270
you accumulate the dependencies dependencies could be external dependencies like your database your
00:04:55.360
certain api's that you consume you certain web hooks that you use a certain
00:05:00.939
technology that you are using but also it could be internal dependencies as the application is growing a lot of modules
00:05:07.629
do you use it inside they are dependent to each other are they dependent in a general way and dependencies only you
00:05:16.030
see this massive dependence and seven massive things and you're trying to fix something and you're trying to like make
00:05:22.120
a new feature you're trying to a Joe that you make certain adjustments you actually created more problems and by
00:05:28.930
creating more problems you end up in a mess like that and this is the mess
00:05:34.680
honestly I don't know how you can work on this like if you need to do something it will take you years to figure out at
00:05:41.560
least what's going on and dependence is
00:05:46.569
often I'm not sure if it will work though yeah looks like it works dependencies and and
00:05:54.190
the work with that is like it's like you're an operator on a moving train your train is moving your application is
00:06:00.580
working you have customers each second thousands of customers each second and you see that you're undertaking a
00:06:06.819
problem in front of you and now you are on a moving train trying to fix the problem and changing the views in a
00:06:14.110
moving train it could be really hard things to solve and it often looks like
00:06:20.319
that and it's enormous stress to work like that so this is one of the carters problem
00:06:26.979
that software engineer could face it's a problem on thread in the needle it's a problem on finding what is your business
00:06:35.379
logic looks like what would when you have a feature what exact feature and what exact place in the application you
00:06:41.800
actually need to change what those where use words mean in your court because court rarely
00:06:49.810
reflect the actual business processes that you application has and that's why
00:06:55.360
you have this and now you need to find what does it mean to adjust the balance
00:07:02.620
how I'm adjusting a balance for my customer I don't know my user well where
00:07:08.800
to find that I don't know but what we want to see is the same amount of features but features that are
00:07:17.010
segregated in a smaller chunks of information and when we know that we
00:07:22.150
need to adjust the balance or we need to make certain certain design decisions we
00:07:27.160
know where to look for those things we know where to find those things and how we can actually adjust it and it's not
00:07:35.200
an easy task of course and it's a task that can lead to you to the vertical
00:07:40.740
architecture where your each individual part of your system is in dedicated part
00:07:46.180
of your codebase it could be even different projects that communicate between each other but I will talk about
00:07:52.600
it a bit later as Alan Kay I heard that if you enter the quote from Alan Kay
00:07:58.330
your presentation is way better so I definitely find Alan Kay quote and alan
00:08:05.020
keyes said one that one of the things that suffer engineers need to do is to
00:08:11.110
find something that is a bit better than you have right now instead of trying to find the silver bullet that you'll
00:08:16.479
definitely solve all of your problems you just see on your application what is the worst thing that you have right now
00:08:21.820
probably the worst thing is the one that you scary the most if you have the permit or implication that you scare
00:08:27.250
about this is the most important thing to solve right now and this the thing that you actually need to to find a
00:08:33.700
better way how to solve it and I want to talk about DDT today and DDT is not the
00:08:40.870
Holy Grail DDT that you know is the one that we use often it's a database driven development and database driven
00:08:48.880
development is what we often do when we start an application and we have certain certain tasks that you need to do
00:08:54.850
something something simple something simple let's say a transfer money and I need to adjust the balance for them for the for the payee and for
00:09:02.970
bar whatever it just I came up with this example anything okay I have I have
00:09:09.030
users and they have a transactions and users probably have a balance so I will create a new table example call it users
00:09:16.320
they have balance whatever it works I will create turns actions transactions
00:09:21.900
connect users and there is a certain amount and now I need to think okay I have two tables how and where I should
00:09:28.740
put my logic my logic did should represent what is actually I'm doing I need to transfer 10% of my balance and
00:09:35.880
my amount to my balance I can do it in a in a MySQL or pause briefly where
00:09:42.270
triggers or procedure it's kind of hard to scale with people maybe I heard this
00:09:47.340
is some people use callbacks and callbacks a silver bullet you can use it everywhere I go just put in a callback
00:09:53.130
each time when we create the transaction I will adjust I will adjust the balance and will adjust the balance for both
00:09:59.160
users it works I made some tests works fine no problems let me have another things that we need
00:10:06.360
to do those transactions that will check negative or positive so we wouldíve you
00:10:11.850
to be able to give them a type just to give it a type to use some report to display it somewhere not a big deal sure
00:10:18.390
I can create a migration and immigration I can make this decision either I need
00:10:23.460
to create it as a as a lawn or the credit base and type and base a balance
00:10:28.550
but I run it in production and suddenly I have so many complaints that balances
00:10:35.130
all over the place for people and I see this little thing that I added to my immigration I saved it and when I saved
00:10:41.640
it I run my callbacks for all existing transactions and when I run all of this exists and transactions I had this and
00:10:50.870
it's a simple example I just made it here but usually we have 10 to 12 to 20
00:10:56.820
callbacks those callbacks on other callbacks and we have a callback hell and come back holes in immigrations and
00:11:03.300
maintenance tasks it's a bad thing to have if you of course can make workarounds
00:11:09.550
lots of people do that but it's not about workarounds in the next example we
00:11:15.240
we have another a part of the application and in this part of vacation we want to make users be able to upload
00:11:22.390
images we have image we have same user they have a name it's a pretty plain
00:11:28.090
string they can come they may make it can make a name for their for their images and then users sad you know what
00:11:36.250
I drew like this image and I want to give it better description so just copy paste from wikipedia the whole text I
00:11:43.090
paste it in the input and it'll try to save it and I cannot and it's a bad
00:11:48.310
experience and as you grow and is it is the application growing as you like in the first steps of your of you growing
00:11:55.260
growing curve those bugs and those problems they are declining the trust
00:12:01.870
that you have with your customers and declining the trans meaning that you will grow will slower and it may end
00:12:07.990
that that you may not grow at all and you may sunset your application so what
00:12:15.820
I mean by that I mean with that but that your business operations and your things that your business is working on and
00:12:21.490
really care about it's not your data base it's not the thing that you that
00:12:27.820
you need to worry about when you design and when you grow your business data base is just an artifact of your
00:12:33.880
business operations when user wants to applaud something it has a starting point and an ending point database and
00:12:41.200
dollar stores it could be not only one database it will be multiple others it's just a legacy it's just the history what
00:12:48.130
happened by the database in the understory you may retrieve the history or what happened but it's not an actual
00:12:54.490
operation that happened so thus it's better to separate your
00:13:00.130
persistence from your business logic you can separate your persistent by many ways in if you use in rails it could be
00:13:08.380
active model or active record and if you use sequel it could be a sequel record
00:13:13.570
and there is lots of other ways how you can you if you don't know if you want to you can even separate it by additional
00:13:19.930
layer that you place in between your application your business logic and your persistence so you end up with something
00:13:27.430
like that and your persistence layer is sheltering you from the database the shelter and your business separation
00:13:33.010
from your database that's all your all your logical things that you need to do
00:13:38.380
all that your business is supposed to do is independent from the way how you
00:13:44.320
store those artifacts and when it's independent from what you have in store reserved tracks you can write your tasks
00:13:50.830
better you can shelter it you can make it faster you can make sure that it actually works and you don't need to
00:13:56.470
worry about how you scale your database and your scaling of your database is not related to your business logic in the
00:14:04.210
next example what would happen if we we have a simple task now we have
00:14:10.900
transaction and we have images and we want to display some analytics about it pretty simple we made up analytics
00:14:18.460
controller and I should call it visible with that but usually it's a big massive
00:14:24.580
SQL and that's how it's usually looks like you you were thinking okay I have a
00:14:29.980
persistence layer somewhere I don't want to put it in anywhere because it's not related to any of my tables and my
00:14:36.310
models let's say we put it just in a controller whatever yeah I will put in a controller is just one small analytical
00:14:42.550
thing I put it there no one will worry I will have some reviewers my friend who can review it and we can approve it and
00:14:48.550
ship it ship it then we have another thing you know the one table that you
00:14:54.160
made there it's really cool we really want to hear it in CSV or XML format because we need to share it with our
00:14:59.800
executive and they need to see how fast we draw in you think yeah sure I can do
00:15:05.500
that I can have a small buttons there let's applaud it down Laura and you should have visible that I made
00:15:11.390
a another slide which is a bit better but it's not really really important how
00:15:16.760
it's written it's the concept that it represents so you did two things they
00:15:21.920
you made it to CSV to XML you wanna filter if it's Excel all you want to filter the empty strings or whatever you
00:15:28.730
need to filter for if sanal xml could be really tricky the really depends on the schema so all of your logics go there
00:15:34.430
because well you haven't made a decision where this logic should go so it go where the where it was made it love last
00:15:40.670
time in a controller and it could be a really big thing that you end up in at the very end it could have like lots and
00:15:47.690
lots of like small conditional not related to your initial task and then
00:15:53.899
your business said you know we have a new awesome marketing tool that you want to have and they really would like to
00:15:59.720
have this data can we send it to them any like not really so this means that
00:16:10.300
ideally you need to segregate you interfaces from your business logic because the moment that you have certain
00:16:17.089
requirement to do like to display certain data in certain format this represents a certain business value that
00:16:23.209
this thing this particular thing is for you for your business and thus
00:16:28.850
separating those interfaces could pays you back once you need to provide this
00:16:33.920
this particular inter business thing to other interfaces like CSV like XML like
00:16:40.160
web hooks so you end up here where you have lots of interfaces that all point
00:16:45.740
in to your business logic I will expand on this a parodic method back and bit
00:16:51.709
farther they later of course I promise that you want to grow and you want to
00:16:57.320
grow fast and it's supposed to be an easy thing but it's not an easy to make it to make
00:17:02.930
it easy because simple is not easy it's really hard to make something simple
00:17:08.890
because you need to invest time and you need to invest your knowledge into making things simpler
00:17:15.540
I will talk about a bit how you can do that how isn't very important important
00:17:23.430
thing because if you don't say how people don't know how to do that then they see the problem but they don't
00:17:29.400
knows the solution in ideal situation that we know who it's not me in ideal
00:17:39.240
situation we want to grow this time and we're going to grow our features and our complexity would be a lie near at least
00:17:45.360
somehow a lie near complexity is the community it's not only a amount of
00:17:52.350
objects that you have in your system it's also a communication between those objects also play a big part in their
00:17:57.630
complexity and thus it usually we end up in a situation like this when the amount
00:18:03.030
of features over time the not allowing us to do anything more than that each
00:18:09.930
new feature takes ages to develop and we don't know why let's stick down to the rabbit hole and
00:18:17.420
see how we can simplify at least a bit simplify this problem for our future and
00:18:24.650
ongoing development and the main thing here is that all other things are super
00:18:31.440
simple they're making web UI JSON interface XML it's a mechanical task
00:18:37.400
filtering XML is a mechanical task making a persistent very easy and extendable in mechanical tasks those
00:18:44.160
little errors on a bottom they basically represent some it may not even exist if
00:18:49.620
you don't need it but it represents the data flow that you need to send from one data store to another one for whatever
00:18:55.440
reason you need for some reporting or from other stuff or some backups those small scripts they're not related to
00:19:02.160
your business they related to your infrastructure and those infrastructure things they could be complex to grow to
00:19:08.940
grow you to grow your infrastructure but if not related to your business directly it's more related to how you roll your
00:19:16.050
application in a form of servers and in form of data flows
00:19:21.410
the juicy part of this diagram is the business logic is the thing that rarely people know how takes how to work with
00:19:29.330
and I remember back in the days everyone everywhere was the fat controllers and
00:19:35.090
thin models and people like oh we have so big controllers it's all over the place it's like some thousand of lions
00:19:41.120
and we didn't know what to do with them then someone came and said you know what thin controller and that model will
00:19:47.000
solve all your problems in the world and you're like oh nice if you've routed everything in a in a model and now and
00:19:53.690
then you see all those callback hell's problems and you see there is certain concepts that intermix with each other
00:19:59.450
and you probably accumulated lots and lots of mix-ins all over the place and you mix in pink things around and you
00:20:05.870
don't know what the objects are representing anymore and then someone came and said you know what everything
00:20:13.370
should be thin but we add in services further to your life and everything became a service even the things that
00:20:18.980
not a service now right at that point became a service if things that represent certain business value certain
00:20:25.549
data flow certain data that you sent between your modules it became a service
00:20:31.730
which is always not true often not true so let's see what your business logic
00:20:37.880
often looks like you probably have few God classes probably one maybe two God
00:20:43.940
class it's the one that's tickling around like thousand or 2,000 lines and one change in this class
00:20:50.600
putting down the whole application and the main architecture and your project says don't go there and don't look back
00:20:57.049
like you don't make any changes there just just don't and you probably you
00:21:03.260
wanna you wanna fix that and you want to change that and that's what I'm currently do each time when you see
00:21:09.890
problem love your problem love your problem and try to understand the problem and then try to fix your
00:21:16.400
problem because this will help other people and it will help you grow with time we can apply certain principles all
00:21:23.929
architecture are software engineering those problems they're not new and the solution to the problem they they're not
00:21:29.690
new so many design see that there is a certain certain
00:21:34.850
parts of the application require your attention unique you can apply first you can apply certain principles to make it
00:21:40.730
easy and clear what's good what's exactly going on in each individual object one of this principle is Hika
00:21:46.610
Hashem the Hessian stands for the understanding of the objects about
00:21:52.490
themselves each object should be meaningful enough to justify the existence but not have
00:21:58.429
lots of meanings inside that each individual part of it will be different one from another so it should represent
00:22:05.870
one idea not many ideas that's what your property project is and if you apply it
00:22:10.880
you got got classes and you got objects they usually represent lots of things they could be they could mix those
00:22:18.650
Keshav ideas together and you need to thread in thread a needle and understand
00:22:24.470
what part is related to to what other part and slice this and slide this big
00:22:30.020
object into smaller pieces and use the smaller pieces where it's important so you end up something like that you see
00:22:35.900
how simple I'm just made a slide and you're done but usually it's of course it's a way harder problem you actually
00:22:42.200
need to dive in you spend time and make sure that you fix in the right thing
00:22:47.799
next principle is you want to make it low couple lower coupling and lower
00:22:53.630
coupling means that you don't have lots of dependencies for your objects it's decoupled from other object as far as it
00:23:01.010
can be because having lots of dependency meaning that you have lots of things that you would that the object is responsible for and it's it's usually a
00:23:08.630
bad thing to have so you implying some decoupling principles and it's still kind of a lot of things to lots of
00:23:16.700
things going on there is a few other things that you can work on and you can apply you can apply other principles one
00:23:24.590
of the died like for example is dependency direction it meaning that when you design in your system you will
00:23:32.299
still have the modules that are more important than others there are things that are really represents some core
00:23:38.330
ideas and the things that are supporting ideas or some objects that the utility
00:23:43.760
classes you need to transform data from one from one type to another one and in this case
00:23:49.429
in this case is your higher-level modules they shouldn't dependent on a lower level modules your lower level
00:23:57.679
modules you can think about it is a duck typing your if you if you family resists idea so your main object is not
00:24:04.849
dependent on your objects that are coming into your main objects those smaller orbitals should be dependent on
00:24:11.539
the interface that your main object is given here and thus you you make a right
00:24:18.199
direction of your dependencies and at this point it still could be a really really messy you can have like thousand
00:24:24.619
of thousand a thousand of objects each of them may be in general wouldn't be hard to work with but as a general
00:24:31.969
concept and general ideas you didn't know what who is responsible in a general way toward other things and
00:24:39.669
that's you need to find and you need to bind your business context you need to
00:24:46.099
find the objects that are coming together and updating together there is
00:24:51.469
lots of techniques from scrolling through github or SVM and finding what objects were coming to were most often
00:24:59.179
changed together so you can find those objects or you can start writing a dependence in app and see what is the
00:25:05.719
what is dependent what kind of dependencies you object has and by finding those business context you can
00:25:12.469
find those objects that are important to what other parts of your system and at this point you can already kind
00:25:19.579
of slow down this is already good enough to start working towards to better
00:25:25.579
things you already know if you see if those on left side rats and in the middle those things are present certain
00:25:32.690
ideas and you and your team probably know that ideally in your application you can have a different different
00:25:38.749
folder for each of those concepts it's still same application but on a
00:25:44.419
conceptual level this application now has different business contacts and each business contacts represent a ad about
00:25:51.440
it it could be like one is analytic context and other one it could be sales contacts third one could be imaginary or
00:25:58.730
something like that the one thing that will work with images that we talked about before but you can go to can go
00:26:04.879
farther and you can separate them in a free different business contacts and through an essentially in three
00:26:10.580
different application but you always need to remember about midcal flaw mid-gulf law says that your your
00:26:18.350
complexity of the system it's not really a correlated to the amount of object
00:26:23.779
that you have but also correlated to the amount of connections that operatives have between themselves it may solve you
00:26:29.690
some problem separating things in a smaller pieces but it will also create you and other problems that comes to
00:26:36.769
them comes with them there are some solutions in the market that are on in
00:26:41.869
our community right now that can help you and it can grow your application this time if it's if you have a small
00:26:49.730
application maybe something like you playing Ruby file will work probably until the thighs 5,000 of time I love
00:26:55.759
lions you can work with it within one file I've seen people done that I didn't like
00:27:01.580
that but it's probably work for some people if it's go more than that it's
00:27:09.169
becoming a huge pain and some other things like Sinatra grape can we can fit
00:27:14.539
in this particular small small in this particular small SEC segment then of
00:27:20.840
course it's really a new application can omit that I did I like and I really like
00:27:26.990
the people diving into that and if it's bigger than that you look in honest
00:27:33.919
solutions that are not rails anymore and it's probably not even her name anymore you trying to grow the concepts rails
00:27:40.610
and Honami and trailblazer and rarbie those are the concepts that are Universal soldiers they are made for 80%
00:27:48.320
of the time but you still have 20% of the time and you still have a solution that you need to build for yourself and
00:27:54.950
where in in this particular case they may fail you and no one in your in
00:28:01.480
no one in the world knows better your application than you are knows better
00:28:07.750
application than you are that's why you have the most contacts and that's why you may build your business logic better
00:28:14.620
because when you evaluate those kudos those additional things those if you want to bring Trailblazer to application
00:28:21.460
or you want to bring drier B to your application you need to think about it as a dependencies and when you think
00:28:26.710
about dependencies if you wanna if you if you go in and I in a supermarket and you want to have something sweet you
00:28:32.740
don't buy a cake right you usually buy quite a small piece of a cake if you buy a huge cake you probably regret about
00:28:39.790
that will you regret about that with time so if you buy in a smoky and don't buy a cake if you want a small piece and
00:28:47.669
ideally in your engineering you need to end up with a situation where the only one condition is the one that you're
00:28:54.280
choosing the object that you are using right now in each particular situation and this the end with the arm partisan
00:29:01.090
port sorry for using without asking and
00:29:07.900
that's it you can ask me question if you have any and I said I love your problems talk to me afterwards
00:29:21.960
questions on the talk or otherwise any guys come on up
00:29:28.140
oh there you go of course Tim are there any books or places that you could
00:29:33.840
recommend for sort of learning that change in mindset I would see lots of
00:29:39.480
DDT books they are really wake you sometime and it's like a big books eight
00:29:45.300
hundred of em of pages but they can help to build their mindset that will think
00:29:51.420
about objects and businesses instead of just throw something in those file and forget about it I like the old books
00:29:58.830
about for pattern guys and girls about patterns of object-oriented suffer lots
00:30:07.020
of sentiments books and seminars of course guard branched in few stalks from
00:30:13.170
gari brush named with Gibbs yeah that's that's it that's a good one that's a lot
00:30:20.280
of resources any more questions yep sure
00:30:26.820
could apply this well
00:30:35.460
quite often your existing application will have a lot of complexity and when
00:30:41.019
you start developing an application you just go with the flow and you put it a lot of pieces together at a lot of
00:30:46.749
requests as they come by and so see you've reached a point in your application where you develop a lot you
00:30:54.549
wanna roll back the weight and stop using this approach to smooth things
00:30:59.559
over so how do you have is this what this particular with this idea apply or
00:31:05.950
Tony kobold yeah those think you have an idea supply ask for things is this application as
00:31:12.549
for new one when you build a new one you it's it's way easier than refactoring
00:31:17.889
the old legacy systems I have factored lots of all legacy systems all of that is it's a big pain but you can soften
00:31:24.849
the pain so you are by applying application a better planet is if you
00:31:30.129
have a legacy system first thing that you need to do is reveal the truth diagram true diagram means that how many
00:31:36.549
times the file was changed and what is the complexity of the file in this diagram you meet an axis and y and see
00:31:44.229
those files that the change in most often and most complex one and you start from them because they are the most
00:31:50.769
problematic one in your application Vietnam and then you slowly peeking in and slowly trying to make this refactor
00:31:57.479
refer green rather factoring I try to make test that's if you have test often
00:32:03.700
you don't have dust that way you need to do this first and you trying to
00:32:09.220
understand the context of you of the things that each particular system does and then eventually fortunately and
00:32:16.239
hopefully you will be able to make it simple
00:32:21.350
did that answer the question I think the LA Convention I mean it's
00:32:31.530
it's it's you know if you want to soften the blow it's that we're gonna take off people or if he referred to you know
00:32:38.600
research the back of your for and see how you can break it down a bit into clusters and stocks are sure
00:32:44.780
step-by-step anymore guys come on don't
00:32:52.260
be shy there you go when do you know that what
00:32:59.960
do you get a feeling that you over have detected application try not to over
00:33:07.249
architect try not to solve problems it doesn't exist if you feel and you see that you you
00:33:13.490
have problems that I exist and if you see that certain features are starting
00:33:19.039
harder to develop and certain things are really and you scared or certain parts of the application that's problems start
00:33:25.759
when you need to do arbiter like when you need to repair before that not try to invent the perfect thing just make it
00:33:32.990
good enough thank you thank you for the
00:33:38.960
question and last question come up with a one
00:33:49.350
all right that's it thank you so much Alex
