What Sucks About Erlang (2008)

My point is that you can achieve many of the same benefits of microservices within a single app (or "vm" in erlang terms).

So you get isolation (beam processes), decoupling (message passing) and resilience (supervisors) without having to deploy and maintain N different apps and the pain points that come with that (monitoring, discovery, integration testing, debugging, ...)

Our company has other systems, in other languages, and a pretty involved deployment process (docker+nomad) so our deploys are pretty specific to that. But, I can tell you that as a developer, having only a few "apps" that internally have hundreds of services, has more or less freed us up of having to worry about any of that cognitive (and performance) overhead.

I don't quite get pouchdb. Do you need to store multiple revisions of the same document in a browser database? Locks/transactions seem like a better approach here for writing to data client side.

It's a 10 year old blog post from the creator of CouchDB who then later declared the project dead despite it already having a vibrant open source community with multiple users and contributors. Then went on to create another project that had a similar name but was a merge of a bunch of other technologies which lead to endless confusion years on after the fact.

http://damienkatz.net/2012/01/the_future_of_couchdb.html

Yes, of course the argument should stand on its own despite the author's other claims or blog posts, so let's review some claims:

> Because Prolog sucks ass for building entire applications.

There have been entire applications written in Prolog. Attacking syntax is is just silly. Had Prolog used statements and not expressions or ; instead of . I doubt we'd would have a seen a lot more or less applications in Prolog.

> But that hasn't deterred Erlang from stealing it's dynamite syntax.

The syntax is consistent. That's the point. It has relatively few syntax and semantic rules. It has strong types and behaves consistently, comparing say, with "this" in Javascript. js has semicolons but it behave inconsistently sometimes. You'd have to explain why {} + [] = 0 for example. Picking on js here since the author used js as an example of a "good" language.

> editing code is much harder than conventional languages.

Have not found this to be true. It is a bit harder. But not much harder and not a show stopper. You can't randomly copy and paste code and reorder statements because compiler will not like it. That's probably a good thing. Variables being immutable is helping when inserting a new function call in the middle, you have to think well if that modified value is suitable for the operations that follow below.

> If Expressions

There are 6 paragraphs talking about the "if" expression. "if" is not even used that often and focusing on it just because it has the same name as the "if" statement from Algol (C) type languages is a waste of time.

Use "case" instead or even better multiple function clause heads. Yes, you can't do an early return so you have to explicitly handle every case. I see it as a good thing. You can cheat and say "_ -> ...DefaultValue" but it will stand and be obvious what's happening to the reader.

> You Say String of Characters, I Say List of Integers

Or list of unicode code points... You have a choice to use binaries instead of lists of integers. They are closer to the traditional strings in other languages. This is a better explanation of how they work: https://medium.com/@jlouis666/erlang-string-handling-7588daa...

> but it's surprisingly limited and verbose, requiring you to state the type of the record for each reference in the code.

Since the blog was written Erlang got maps other languages call them dicts or hash tables for example.

All in all I'd say syntax is often what people talk about but it is closer to bike shedding. It is easy to argue about because it's right there and everyone can see a . instead of a ;. It would better to talk about the semantics of the language, sending messages, concurrency, modeling applications as a set of communicating processes, supervision trees etc.

I disagreed up until the docs section.

I like all the other things he critiqued!!1!

Rearranging: the punctuation helps you think about how you’re rearranging. You can’t copy/paste around in any function and expect it it work!

The records comparison to a struct is unfair. He could have patterned matched on the record instead of destructuring and accessed the attributes of the record.

Immutability! Fuck yeah! I tend to name subsequent variables with the mutation that was applied to them. It’s wordy.

Name = “wayne”

NameCapitalized = ...

Now get off my lawn!

It doesn't use real node-local/shard-local MVCC AFAICT. It's O_APPEND on flat files.