The source content for blog.juliobiason.me
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

560 lines
32 KiB

<!DOCTYPE html>
<html lang="en">
<head>
<meta http-equiv="X-UA-Compatible" content="IE=edge">
<meta http-equiv="content-type" content="text/html; charset=utf-8">
<!-- Enable responsiveness on mobile devices-->
<!-- viewport-fit=cover is to support iPhone X rounded corners and notch in landscape-->
<meta name="viewport" content="width=device-width, initial-scale=1.0, maximum-scale=1, viewport-fit=cover">
<title>Julio Biason .Me 4.3</title>
<!-- CSS -->
<link rel="stylesheet" href="https://blog.juliobiason.me/print.css" media="print">
<link rel="stylesheet" href="https://blog.juliobiason.me/poole.css">
<link rel="stylesheet" href="https://blog.juliobiason.me/hyde.css">
<link rel="stylesheet" href="https://fonts.googleapis.com/css?family=PT+Sans:400,400italic,700|Abril+Fatface">
</head>
<body class=" ">
<div class="sidebar">
<div class="container sidebar-sticky">
<div class="sidebar-about">
<a href="https:&#x2F;&#x2F;blog.juliobiason.me"><h1>Julio Biason .Me 4.3</h1></a>
<p class="lead">Old school dev living in a 2.0 dev world</p>
</div>
<ul class="sidebar-nav">
<li class="sidebar-nav-item"><a href="&#x2F;">English</a></li>
<li class="sidebar-nav-item"><a href="&#x2F;pt">Português</a></li>
<li class="sidebar-nav-item"><a href="&#x2F;tags">Tags (EN)</a></li>
<li class="sidebar-nav-item"><a href="&#x2F;pt&#x2F;tags">Tags (PT)</a></li>
</ul>
</div>
</div>
<div class="content container">
<div class="post">
<h1 class="post-title">Why Rust and not Go</h1>
<span class="post-date">
2019-09-16
<a href="https://blog.juliobiason.me/tags/go/">#go</a>
<a href="https://blog.juliobiason.me/tags/rust/">#rust</a>
<a href="https://blog.juliobiason.me/tags/programming-languages/">#programming languages</a>
<a href="https://blog.juliobiason.me/tags/nitpick/">#nitpick</a>
</span>
<div style="border:1px solid grey; margin:7px; padding: 7px">
<p>This is a rebuttal to <a href="https://kristoff.it/blog/why-go-and-not-rust/">Why Go and not
Rust?</a>.</p>
</div>
<h1 id="huge-disclaimer">HUGE DISCLAIMER</h1>
<p>Before jumping into the discussion, let me put this first: I do write code in
Rust (not professionally), I've not written something in Go yet but I keep
reading about the language and its ecosystem. I also follow the Rust as a
language and its ecosystem.</p>
<p>Second, another thing you must know about me: I've been a developer
professionally for about 30 years (its no hyperbole here: I did start
professionally writing code when I was 12 and didn't leave the field yet).
I've written code that run in about 15 different languages, so I have strong
opinions about coding after suffering with those languages.</p>
<p>Third, I do believe languages do not exist in a vacuum: Besides the language,
you have libraries and frameworks; besides the languages and the frameworks,
there is dependency control; besides dependency control, there is a community.</p>
<p>Fourth, if that's not really clear so far, all this is a matter of opinion --
even the original post is, although it doesn't say so.</p>
<p>Fifth, yes, I did take things out of order, mostly 'cause I thought some
points are scattered around the original text that are complementary.</p>
<p>That being said...</p>
<h2 id="the-nitpicking">The Nitpicking</h2>
<p>(Those are points that just brushed me in the wrong way, but they are not
major points and you can really skip this if you don't want to read a bunch of
complains.)</p>
<blockquote>
<p>Go is fast, but Rust is faster.</p>
</blockquote>
<p>C is even faster. So what?</p>
<blockquote>
<p>Go has an efficient garbage collector, but Rust has static memory management.</p>
</blockquote>
<p>And so does C and C++. So what?</p>
<blockquote>
<p>Go has interfaces, but Rust has traits and other zero-cost abstractions</p>
</blockquote>
<p>Well, Go doesn't have interfaces <em>per-se</em>. Their interface is akin to Python
&quot;if it has an <code>open()</code>, <code>read()</code> and <code>fseek()</code>, then it's a File-like object&quot;,
as far as I know. Python even changed that from the &quot;it looks like a
file-like&quot; to using the double-dunder functions to fit the interface (&quot;if it
has a <code>__len__()</code> method, it is an object with a size&quot;).</p>
<blockquote>
<p>Go has great support for HTTP and related protocols and it doesn't take long
to write a satisfactory web service.</p>
</blockquote>
<p>More satisfactory than <a href="https://palletsprojects.com/p/flask/">Flask</a>, in which
you can create a service (a very dumb one, in that) with only 5 lines of code?
Does it? Or is it a <em>personal opinion</em>?</p>
<p>Honestly, I haven't seen -- even with Rust -- something as dead simple as
Flask, so there we have it. If you need performance, one could use
<a href="https://github.com/huge-success/sanic">Sanic</a>, which is a uvloop powered
server with a syntax that is pretty close to Flask.</p>
<p>So, when you say &quot;great support&quot; and &quot;satisfactory&quot; is that a <em>fact</em> or an
<em>opinion</em>?</p>
<p>Obviously it is an opinion, as much as me saying Flask/Sanic can beat anything
Go has, and neither of those are valid for anything. Beauty is in the eye of
the beholder.</p>
<p>(But if I'm nitpicking, I can throw whatever language I want here.)</p>
<blockquote>
<p>The creators of Go like to call it a “boring” language.</p>
</blockquote>
<p>Weirdly enough, I heard the same thing about Rust. So which one is the
&quot;boriest&quot; of them all?</p>
<p>This kind of call is akin to the PyPy devs saying that &quot;PyPy is 300% faster
than CPython -- for tests written specifically to prove that PyPy is 300% faster
than CPython&quot;. The same thing can be said here: Go/Rust devs call their
language boring 'cause they want to prove their language is boring. Neither is
true -- and, weirdly enough, <em>both</em> are true.</p>
<p>One could even claim that Python is more boring that Go. <sup class="footnote-reference"><a href="#1">1</a></sup> And Rust, by the
way.</p>
<blockquote>
<p>do more with less” has proven to be very successful.</p>
</blockquote>
<p>Wait, are you <em>really</em> quoting Dennis Ritchie, in which he was defending C? C
has less than Go or Rust, so maybe we should jump back to C? Is that what you
mean?</p>
<blockquote>
<p>In truth, none of these things alone is particularly impressive, but they do
describe the mindset that Go wants to impose. Many don’t like it but, in my
opinion, it’s a killer feature for some types of development, like
enterprise software.</p>
</blockquote>
<p>If we point that Rust does exactly what you're saying and <em>better</em> (&quot;enforce&quot;
vs &quot;you need to run something, otherwise nothing changes&quot;, which I <em>will</em>
discuss later in this post<sup class="footnote-reference"><a href="#2">2</a></sup>), then Rust kills Go in enterprise software, right?</p>
<blockquote>
<p>Enterprise software always has a big scope.</p>
</blockquote>
<p>Yes, and that's why we break this scope -- and some domains and everything
else -- into smaller parts that connect to each other. Those are called
&quot;microservices&quot; and one thing is that you can write them in whatever, but the
scope is always a small one (for different scales of &quot;small&quot;) but it won't
be something ginourmous like a monolith, in which all the scopes exist in the
same place.</p>
<p>(I'll make this point again later, but it is weird how the author says one can
spin a Go app really fast, but then comes with the sort of points like this
that make sense only for monoliths, and I'm not sure which is the real point
being pointed in these cases.)</p>
<blockquote>
<p>To unravel complex domains you need a well-structured process.</p>
</blockquote>
<p>(And then he jumps into discussing about domains and expects and stakeholders)
You know what you're talking about? A DSL. You want a DSL to is close to the
experts about their domain, you want a DSL so stakeholders can understand
what's going on above the code.</p>
<p>You want Racket. I've seen things close to this in Rust using macros (which is
witchcraft to me) but I'll refrain from saying &quot;Rust has it, and better&quot;.</p>
<h2 id="the-freaking-cargo-cult">The Freaking Cargo Cult</h2>
<blockquote>
<p>Go was created at Google to solve Google problems</p>
</blockquote>
<p>This is something that we, developers, who love and hate <em>any</em> language need
to discuss. And it comes in this other point:</p>
<blockquote>
<p>As I already mentioned, Go was created to solve Google problems, and Google
problems are definitely enterprise-scale problems.</p>
</blockquote>
<p>You know who has Google problems? GOOGLE! You know who else has Google
problems? NO ONE!</p>
<p>No one is a huge search engine that lives capturing peoples data to provide
relevant ads (and, sometimes, search results, and shopping lists and
whatever).</p>
<p>It's the same bullshit people claiming &quot;Netflix has 600+ microservices using
Spring, so we should use Spring for our microservices&quot;, 'cause you won't have
600+ microservices, and most probably none of them are related to providing
video streaming.</p>
<p>It's the same bullshit people claiming &quot;Amazon deploys a microservice every 11
seconds, so we must use microservices too!&quot; 'cause, again, you're not a huge
cloud provider with two or three different versions of the same solution.</p>
<p>This &quot;Cargo Cult&quot;: The idea that if we do the same thing someone else did and
it worked, it will work for us too. Enterprises everywhere run Java, for huge
scales -- AWS is the first that comes to mind -- but you're not claiming Java
can solve &quot;enterprise-scale&quot; problems, are you? The point that you're trying
to make here is &quot;worked for Google, will work for you&quot;, which is plain wrong.</p>
<p>And it doesn't even involve Go or Rust. You're trying to make a point by
saying &quot;They use&quot; and that's not a point. <em>At all</em>.</p>
<h2 id="the-plain-wrong">The Plain Wrong</h2>
<blockquote>
<p>Go is also strict about things that other languages are usually more lax about.</p>
</blockquote>
<p>Guess you never saw the borrow checker. Or the fact that Rust doesn't allow
passing an u32 (unsigned int of 32 bits) as a parameter that requires an u64,
even if the later is larger than the former. There is no implicit conversion
in Rust and a Rust dev must explicit convert from one type to the other.
That's very strict in my opinion, even if you believe this is just an
annoyance.</p>
<p>You can't have strictness with flexibility. The are opposite points: Either
the language is flexible (allowing you to use an u32 in a u64 parameter and
doing the conversion to you, for example) or it's very strict (like Rust
does).</p>
<p>Also, since you're talking about &quot;strictness&quot;, let me ask you this: Have the
Go core devs fixed this?</p>
<pre style="background-color:#2b303b;color:#c0c5ce;"><code><span>result, err := some_function()
</span><span>return result
</span></code></pre>
<p>What is there to fix? No, it's not the &quot;it should complain that err is never
used&quot; (I could replace <code>err</code> with <code>_</code> and the error would still be there). It
is <strong>the freaking error treatment</strong>! You can't call a language strict if, in
2019, it let this kind of stuff slip by. We learnt, in those last years, that
the &quot;not-happy&quot; path happens more often than the happy path. And ignoring such
errors is the major cause of headaches we get, and that's why we have those
stupid &quot;restart job at midnight&quot; cronjobs or watchdogs that keep checking if
the process is running and start it again if it crashes. </p>
<p>I can also bring the borrow checker back into this: You see, we don't talk
about processor speed anymore these days, we talk about cores. The future (and
the present) are multi-thread. I won't deny that launching a concurrent,
multi-threaded service in Go is a lot simpler than Rust, but Go doesn't have
any validation about the memory usage; it won't prevent you from doing
something like sending a structure over a channel and changing that structure
<em>in the same thread</em>. So, while Go makes it easier, Rust makes sure you're not
shooting yourself in the foot in the long run -- which would require the
cronjobs or watchdogs.</p>
<blockquote>
<p>Go doesn’t want unused variables or imports</p>
</blockquote>
<p>Neither does Rust -- it throws a large, explicative warning right in the
middle of your screen -- , so what's the point of this here?</p>
<blockquote>
<p>files belonging to different packages in the same directory</p>
</blockquote>
<p>Hey, do you know that, in Rust, directories <em>are</em> packages, and so you <em>can't</em>
put files of different packages in the different directories? In this case,
Rust seem more strict than Go, doesn't it?</p>
<blockquote>
<p>Go also doesn’t want any “fingerprints” in the code, so it enforces a
single, universal style via <code>go fmt</code>.</p>
</blockquote>
<p>It doesn't <em>enforce</em> if you have to run this. Rust also have a <code>rustfmt</code>, if
that's your point but, besides that, the Rust compiler will complain about
things that are not following the coding style. Try to name a variable in
camelCase and see what the rust compiler will say.</p>
<p>Besides <code>rustfmt</code>, Python has Black (which I hate, but still); before Black,
Python have flake8 and pylint, both which would &quot;enforce&quot; the Python style. So
this is, again, a moot point.</p>
<blockquote>
<p>The toolchain is very often lousy and/or dated.</p>
</blockquote>
<p>You mean, projects don't have <a href="https://rustup.rs/">rustup</a>, the way Rust has?
Crazy!</p>
<p>Or even the backwards compatibility, like when Rust changed from the 2015
edition to 2018, but you could select which edition (compiler version, AST,
and so on) your project would use?</p>
<blockquote>
<p>The Go compiler is fast.</p>
</blockquote>
<p>Ah crap, not that shit again.</p>
<p>The whole point is &quot;compiler is fast, tests run faster&quot;. Well, what if I said
the compiler would catch bugs <em>before</em> the tests? That would be even faster,
'cause then you can focus your tests on system behaviour, which is way more
important than function behaviour or class/structure/module behaviour.</p>
<p>We are, once again, discussing <a href="https://www.youtube.com/watch?v=RAxiiRPHS9k">Fast Test, Slow
Test</a>, aren't we? Let me write a
test for every single function, every single class and oh, look how fast they
run, 'cause the compiler is fast! Then we put the &quot;integration tests&quot; in the
CI and everybody is happy.</p>
<p>Except you wrote tests twice when the only tests that matter are the ones that
check the system behaviours.</p>
<blockquote>
<p>With Go, it’s easier as a junior developer to be more productive</p>
</blockquote>
<p>[Citation needed]. I know this is pretty close a nitpick and I can understand
where this is going, but my guess is that, in the long run, when juniors
understand why certain snippets don't compile, they can be more productive
'cause errors in their code will be caught way earlier in compilation (see
point below about types and above about compiler strictness).</p>
<p>Also, feel free to call [Citation needed] about my point here too, 'cause we
both know we are both pulling data out of our asses.</p>
<h2 id="the-somewhat-right">The Somewhat Right</h2>
<blockquote>
<p>There are a lot of junior developers</p>
</blockquote>
<p>Yes, there are. Also, they are, sadly, not getting that many jobs, 'cause as
this point, nobody is hiring juniors. We can go back to the point of &quot;juniors
being more productive&quot; and say &quot;Hey, junior dev, if you learn programming
language X, you'd get a job, 'cause it's really an easy language&quot; and I'd
throw Python here and break your engine.</p>
<p>Sure, there <em>may</em> be that Go is simpler than Rust (and I'll outright say that
Python <em>is</em> easier than Rust), but we can't say Go is easier than Rust.</p>
<p>If I can go into anecdote mode, I could say that I personally find Go code
harder to understand than Rust, and no, it's not because I've wrote some Rust
code an no code in Go; Go syntax simply does look weird to me, and one can say
that is because the order of the languages I learnt. So, for me, it's harder
to learn Go due its syntax than it is to learn Rust. <sup class="footnote-reference"><a href="#3">3</a></sup> Does the same apply
to junior devs? I'm not sure, but languages are not hard per-se, they may have
some weird stuff, but if you describe the building blocks in order, then <em>any</em>
language is easy to learn.</p>
<p><sup class="footnote-reference"><a href="#4">4</a></sup> Also, let me point out that Rust have an &quot;Ergonomics Working Group&quot; (now
simply called <a href="https://github.com/rust-lang/rfcs">&quot;Language Changes RFCs&quot;</a>,
which discuss ways to make the language easier to use and there is a whole
<a href="https://github.com/rust-lang/rust/labels/WG-compiler-errors">&quot;Compiler Errors Working
Group&quot;</a> which
discuss how to make the error messages in the compiler easier to understand.
With all that, if you junior devs are having issues with the language, they
are invited to go there and show the problems they are having with the
language -- and the Rust community is really focused on that, and all they
need is a Github account, instead of being part of the cabal that decides the
changes. Do Go devs have an open, transparent discussion like that?</p>
<blockquote>
<p>This pushes further down technological concerns such as efficiency, and even
correctness. Don’t get me wrong, the business does care about correctness,
but they have a different definition for it. When you’re thinking about
algorithmic correctness, they are thinking about a reconciliation
back-office for the operations team they keep in a country where labor is
cheap.</p>
</blockquote>
<p>Wait, so it is technological but it is not technological? I know, this should
be in the nitpick section, but there is another important point here.</p>
<p>Thing is, business people do not care about reconciliation; they worry about
deliveries and cheap labor <em>and that is</em>. Are they delivering? Are they cheap?
Good, case closed. We don't care if there is a problem that will appear in 10
months or if they connection is slow between services; it works right now and we
saved money right now, so the math is solid.</p>
<p>On the other hand, if we are talking about &quot;technological efficiency&quot;, you've
already said Rust is faster than Go, so it's more efficient (for the level of
efficient I want to use to prove that Rust is better than Go -- and I'm being
sarcastic here); if we are talking about &quot;technological correctness&quot;, we can
go down the rabbit hole of Rust types and that, although not close to Haskell
types, it forces a good bunch of correctness in your processes. And nothing
about Go types (and their correctness) is ever mentioned; wanna guess why?</p>
<blockquote>
<p>Software projects quickly become huge and complex for all the wrong reasons.</p>
</blockquote>
<p>I have to agree with this. Yes, software grows beyond maintainability and
domains change.</p>
<p>On the other hand, we keep pushing microservices in those larger contexts,
specially to avoid being stuck in certain domains, 'cause you can just
rewrite services (yes, you can) or you break code into different services so
they don't go being the maintainability barrier.</p>
<p>(This point is also annoying the heck out of me, with the comparison of Go
being easy to spin something and then calling monolith problems. Either it is
a language consistent to huge deployments, like Java, or it is a language with
prospects towards simpler things, like Python. There are drawbacks in both and
one can't push towards the to points without breaking everything.)</p>
<blockquote>
<p>Go is much easier to learn than Java or C#.</p>
</blockquote>
<p>Says who? I mean, I don't want to bring the &quot;That's, like, your opinion, man&quot;
card again, but I guess one could find a lot more resources about Java or C#
than Go.</p>
<p>What about Rust? I give that the language is not that easy to pick, but every
working group on the Rust community writes their own book, so one could bring
those juniors devs into reading the books they will need to use and be done
with that.</p>
<blockquote>
<p>The Go community regards as anti-patterns many abstractions regularly
employed by Java / C#</p>
</blockquote>
<p>And I guess Java and C# regards Go abstraction as anti-patterns too, so what's
the point? Any language that has different ways to express abstractions
compared to other languages, and they will call the other language
abstractions &quot;anti-patterns&quot;.</p>
<p>Rust will call some Java patterns anti-patterns too and I have to, once again,
ask &quot;So what?&quot; Does this makes Rust better than Java if it does? Does it make
better than Go if it calls the same Java abstractions &quot;anti-patterns&quot;, but has
better performance than Go and is more strict than Go?</p>
<blockquote>
<p>Go is faster than Java</p>
</blockquote>
<p><a href="https://benchmarksgame-team.pages.debian.net/benchmarksgame/fastest/go.html">For a very small
margin</a>.</p>
<p>But I have to pull the <a href="https://en.wikipedia.org/wiki/Moving_the_goalposts">&quot;moving
goalpost&quot;</a> card in the
blog post here: &quot;Go is simple so that all of this can hold true when
confronting the average Go program with the average Java / C# program.&quot; So now
we are comparing the &quot;average&quot; Go vs the &quot;average&quot; Java/C# (which,
surprisingly, are nothing like the &quot;average&quot; Rust program). Except whatever is
an &quot;average&quot; Go/Java/C# program is never defined, so we can keep pulling data
from our asses and keep saying that the benchmark game isn't valid 'cause the
examples are not &quot;average&quot;.</p>
<p>And whatever I can point as &quot;average&quot;, you will point that is not &quot;average&quot;,
right? That's what a &quot;moving goalpost&quot; means.</p>
<p>Oh, you mean, &quot;on all the things, Go is faster&quot;. As in &quot;it compiles faster&quot;,
which we know means nothing, if it doesn't bring the strictness and
correctness factors into it (let me write a very fast compiler in Bash that
produces code that never runs, but it is <em>fast</em>!); as in &quot;We can fix things
faster&quot;, in which I can call that Rust is pretty close to Elm, in which &quot;if it
compiles, it will run without runtime exceptions&quot;, so in actuality, there are
less bugs (personal observation, it actually does!).</p>
<p>As &quot;average&quot; as in what, actually? As in &quot;whatever point I want to make about
Go being better than Java/C#&quot;, for absolutely no reason?</p>
<h2 id="dropping-all-euphemisms"><sup class="footnote-reference"><a href="#5">5</a></sup> Dropping All Euphemisms</h2>
<p>There one point that, if you take the birds-eye view of both posts, is truly
being discussed here: TTSIW (Time To See It Working).</p>
<p>I'll give it that TTSIW for Go is smaller than Rust, and that seeing something
&quot;working&quot; is a huge step in producing serotonin in your system.</p>
<p>Rust, with all its checkers and rules and strictness have a larger TTSIW,
which may cause the release of glucocorticoid -- the stress hormone.</p>
<p>So, the whole point the original post tries to make is &quot;Go gives me
serotonin&quot;.</p>
<p>But after you are in the field for some time, when you have to go back to the
office 'cause some service is not working, when you have to stop your leisure
time to find out why something is breaking... That's a glucocorticoid release
larger than fighting your compiler for 2 hours. In other words: Having to deal
with stupid mistakes you left behind 'cause your compiler didn't warn you
about them is more stressful than the ones it warned you about, 'cause it is
<em>right</em> <em>there</em>, in front of you, when you're building the system.</p>
<p>I had a talk with an embedded developer about a month ago. His company was
replacing Go applications with Rust, simply because the thing would run for
awhile and then crash because there was some lock he should've used and the
compiler never told him about it. Replacing with Rust was trading the long
term glucocorticoid with a short term glucocorticoid and long term serotonin.</p>
<p>And <em>that's</em> what we are talking in those two articles: Do you want a short
term pleasure and a long term stress or a short term stress and long term
pleasure?</p>
<h2 id="the-things-we-don-t-talk-about">The Things We Don't Talk About</h2>
<p>There is one important piece that is never discussed: Ecosystem.</p>
<p>And no, ecosystem is not simply the number of libraries and packages in the
package manager; it is way beyond that: It's about its community and the way
the management deals with it.</p>
<p>And we have to talk about <code>go dep</code>.</p>
<p><code>go dep</code>, the Go dependency tool, is a replacement of the dependency tool
created by the community, <code>godep</code>, after a whole year with said community
asking for a decent dependency tool, specially compared to the <code>vendor</code>
solution. So, without every inquiring the community, the Go core devs decided
they know better, made a tool and gave a big &quot;screw you&quot; to the community.</p>
<p>Not only that, but just recently the same tool decided to <a href="https://codeengineered.com/blog/2019/go-mod-proxy-psa/">call
home</a> by adding a
proxy on the call of every package, <em>including your private ones</em>. Why? No
real explanation. It simply does. It's not for CDN, 'cause it is just a proxy.
It's not for CDN, 'cause other it would require coordination between the
package repository and the CDN and none of this is included in this change.
Simply, every single install of a package will be captured by Google. For. No.
Good. Reason.</p>
<p>Not only the situation of the core Go devs going against the wishes of the
community, there was even some whisper about forking Go into a community
version, so it could run with a core group that would actually <em>listen</em> to the
community.</p>
<p>And Go is just one year older than Rust. And nobody is saying &quot;Let's fork
Rust&quot; -- even if you count <a href="https://twitter.com/withoutboats/">without boats</a>
comments about a simpler version of Rust, in which he was talking about
<em>another language</em> that would borrow some stuff from Rust, the same way Rust
borrowed some stuff from OCaml and other ML languages.</p>
<p><a href="https://doc.rust-lang.org/cargo/">Cargo</a> is the Rust counterpart of <code>go dep</code>.
Cargo was born in the Rust infancy and it is evolving along the Rust compiler.
All discussions about it <a href="https://github.com/rust-lang/cargo/issues/">are done in the
open</a> with the community input.
Rust itself goes <a href="https://github.com/rust-lang/rust/issues">through the same
motions</a> -- and that could be the
reason async/await is almost a whole year in the cooking, with discussions
about its syntax going through iterations over the issue list.</p>
<h1 id="you-have-to-have-a-posture">You Have To Have a Posture</h1>
<p>You, the reader, may now be wondering why I brought the <code>go dep</code> discussion
into this. You may believe that discussions in the open (and taking longer
than they needed) is a big turn off for you 'cause it makes things move
slower.</p>
<p>But let me ask you this: Which one of those models follow an open source
development model? Let's take Mike Hoye, from Mozilla, definition of open
source: &quot;I think that openness as a practice – not just code you can fork but
the transparency and accessibility of the development process.&quot; </p>
<p>With that, take a step back and re-read the last point again. Which one of
those are really an open project? Is the Go development transparent and
accessible?</p>
<h1 id="in-closing">In Closing</h1>
<p>I have to call the original post completely baloney, mostly 'cause I want this
post to end in a lighter mood. I mean, there is a huge confusion of saying &quot;Go
is better than Java/C#&quot; while what we are talking about is &quot;Go vs Rust&quot;. The
whole &quot;Let me take a huge turn here, saying Go is better than Java/C# only to,
in the end, say that Go is for Java and Rust is for C++, but I'll never
compare Java vs C++ to actually make a point about Go vs Rust with their
comparative other languages&quot;.</p>
<p>And then, when we pick the points in which the author goes straight for the
&quot;Go vs Rust&quot; discussion, all the points are wrong or seemed more about opinion
than actual facts.</p>
<p>So what is actually the point? That Go is better than Java and Rust is better
than C++ and, thus, Go is better than Rust? 'Cause I can totally buy in the
first part, but the second is bullshit.</p>
<p>I won't say Go is a bad language, but Go is in no way a better language than
Rust, specially if we consider the future, in which more cores will be
available (considering the current trends) and more threaded applications will
be more common. I'll say that Rust design decisions give an edge over Go, so
Go should seriously go back to the basics of error control and memory
protection if it wants to be a language for the future. But being bad managed
probably means it would never happen.</p>
<p>And those points make Go a bad option for anyone writing something serious. </p>
<p>But what do I meant by &quot;badly managed&quot;? Well, as you can see, the core devs
don't seem to listen the community on the big issues (one could bring the
discussion about a <code>Try</code> operator, which would be a minor change, compared to
the dependency control, which the community said no and the core devs agreed
but, again, that's a minor thing compared to dependency control tool). The <code>go dep</code> was a complete &quot;ignore whatever the community built, 'cause we know
better&quot; and the Google proxy was simple a PSA, not a &quot;let's ask the community
what they think about it before doing it so&quot; are two signs that they don't
care about what the community wants and that means they can pivot the language
in a way that the community <em>doesn't</em> want and there would be nothing one
could do.</p>
<p>&quot;It's just one tool, not the whole thing!&quot;, you're screaming at me. But it is
a tool the community seriously wanted (the Go Evangelist inside Microsoft came
to her twitter account to loudly say &quot;Go core, fix this or I will stop
promoting go inside Microsoft, 'cause it's bad right now&quot;) and they were first
ignored and then pushed aside.</p>
<p>That's not how open source projects should move about.</p>
<p>You may not care about this &quot;open source&quot; thingy, and that's ok. You may
believe that the core team knows better how Go should move forward than the
people actually writing code in Go, and that's ok. But if you're an open
source proponent, evangelist or admirer, there is absolutely no reason to
defend Go on <em>any</em> accounts.</p>
<p>Want to bash me? Agree with any of the points? Disagree? Come talk to me <a href="https://functional.cafe/@juliobiason/102804921786785051">on
Functional Cafe</a>.</p>
<hr />
<div class="footnote-definition" id="1"><sup class="footnote-definition-label">1</sup>
<p>Added in 2019.09.17.</p>
</div>
<div class="footnote-definition" id="2"><sup class="footnote-definition-label">2</sup>
<p>Update in 2019.09.17, pointing that the discussion about this will come
later in the post.</p>
</div>
<div class="footnote-definition" id="3"><sup class="footnote-definition-label">3</sup>
<p>Added in 2019.09.17, to point that senior devs discussing which language
is easier or harder to teach to juniors is a moot point.</p>
</div>
<div class="footnote-definition" id="4"><sup class="footnote-definition-label">4</sup>
<p>Added in 2019.09.17, pointing that even if senior devs discussing if a
language is harder or easier for juniors to understand, Rust have an open
door policy for this kind of stuff.</p>
</div>
<div class="footnote-definition" id="5"><sup class="footnote-definition-label">5</sup>
<p>Added in 2019.09.17, when I realized what we were really talking about.</p>
</div>
</div>
</div>
</body>
</html>