[erlang-questions] Time for OTP to be Renamed?
Thu Feb 13 19:41:59 CET 2014
Again, I'd be happy to see that changing the name has a positive effect.
For business, the biggest argument, I believe is indeed hiring before
the name. There are not that many Erlang devs -- even your employer, EA,
used to have an Erlang team, and as far as I know and heard from rumors,
they shut down the team after having hiring problems, but this is
third-hand information. Businesses that are willing to allow remote
workers have more of a chance, but to be able to hire Erlang devs
locally is quite a challenge. Nevermind finding an ops team ready to
support the language and its platform.
Back when I started Erlang, I thought the problem was accessibility, so
I set up to work on LYSE. It probably helped a bit, but didn't solve all
of it, far from it.
Before then, the problem was having any book at all, conferences, a
place for the community, etc. These have all been added. Then it was
about how open (or little so) Erlang was, and it moved to github and
then started accepting pull requests. That brought some speed to the
language still, but hey, no popularity problem solved. We organized
Spawnfests (coding contests) two years in a row to try and help until
our team of volunteers ran out of time. That was okay, but not the whole
At some point the issue was needing a killer app. Eventually Erlang got
to power facebook chat, EA's WorldPoker tournament software, CouchDB
(which was even shipped with all Ubuntu CDs for a while, and on the
BlackBerry SDK), Ubuntu One, Riak, Amazon software (SimpleDB), Heroku's
router, videogame chat systems in at least 3 companies, Telecom systems
for phone calls and SMS, real-time bidding stacks for advertisers
(there's easily a dozen of them doing it), DNS servers, Whatsapp
servers, and it's used for internal projects at Apple, and Klarna is a
European bank that runs Erlang at its core, and Goldman Sachs has had
long trials after some of the Erlang code they ran had left the company
in the hands of employees.
Turns out that even though all these corporations and apps have Erlang
inside, that doesn't boost up adoption majorly either.
There have also been talk about IDEs (which were never good enough),
build tools (which were never good enough), more accessible
documentation and other things (which were also never good enough, I
My point ultimately is that yes, the OTP name isn't really great. And it
would probably help a few newcomers to have something that doesn't
mention 'Telecom'. I definitely agree there.
I don't think it's what limited the adoption in big corporations. It
actually entered a crapload of big corporation and keeps getting in more
of them. For the parts of these companies it's ingrained in, it's
actually very hard to get rid of it because it does a good job and is
cheap to operate.
The problem has been in growing the usage within the confines of these
corporations past the skunkworks. Erlang has been used in skunkworks
projects or extremely domain-specific projects for a long, long time,
but hasn't readily made it out of there.
While it would help, I somehow just don't think the 'T' in 'OTP' is all
that significant when it comes to management and staffing. I think what
you're likely to hear is really "But we don't have developers using
Erlang. Who is going to work on that? Can't we just use
<LanguageWeCurrentlyUse> instead? That would be much simpler", nevermind
the operations teams not always willing to go and deal with Erlang
deploys, whether they are live or not.
I encourage and wish the best of luck to the people who'll do the work
in renaming OTP whenever that happens. I'll update LYSE's website
accordingly if it gets in and hope for the best, but I won't hold my
breath. Any progress is good progress, however, and I welcome it.
On 02/13, kraythe . wrote:
> I have read portions of your book and appreciate your insight. However, I
> think you underestimate the task here. Convincing developers may be
> difficult, but if they are good devs they might come around. Convincing
> management with control over budget and staffing when the naming is wrong?
> Nearly impossible. Thats why massive advertising companies have made
> billions off of just naming things correctly. All of the other concerns you
> posted are very legit and I have had and still do have many of them myself.
> But those concerns are at the tech level and only of minor interest to the
> manager wondering why would he staff for erlang and not scala or ruby?
> *Robert Simmons Jr. MSc. - Lead Java Architect @ EA*
> *Author of: Hardcore Java (2003) and Maintainable Java (2012)*
> *LinkedIn: **http://www.linkedin.com/pub/robert-simmons/40/852/a39
> On Thu, Feb 13, 2014 at 12:05 PM, Fred Hebert <mononcqc@REDACTED> wrote:
> > Answers inline.
> > On 02/13, kraythe . wrote:
> > > I Guess my answers would be:
> > > 0) If there is a business case, you can convince them. Low adoption hurts
> > > their maintainability and staffing much more than it does for the startup
> > > or small company. They are a business, not a bunch of unreasonable oafs.
> > That may be doable. I'm not saying the opposite.
> > > 1) Why rewrite the libs if you use the same initials. I wouldn't worry
> > > about that. The programming world is replete with examples of such
> > things.
> > If we can use the same initials, then that's gained and removes a bunch
> > of issues.
> > > 2) and updating the docs will take ... 10 man hours? Do we not have
> > search
> > > and replace capable tools?
> > > 3) Same answer as 2.
> > Yes, but we do not have administration rights to mirrors, say
> > http://erldocs.com/ and translations that can be hosted by the
> > community.
> > The work done with the OTP documentation goes further than the OTP team
> > itself.
> > > 4) Dont need to "make sure" of anything. If the books want to be accurate
> > > they will use the new name, if not "shrug" thats their problem. Trust me
> > > someone on amazon will post "Its not called Open Telecom Platform since
> > > 2014, it stands for "Open Technology Platform". There are enough
> > pedantic,
> > > basement living, people on the internet that will annoy authors into
> > > submission.
> > That doesn't sound like a pleasant experience for everyone. Again, it's
> > not an insurmountable challenge. It's just one more challenge.
> > > 5) Small matter of documentation. "It used to be called X but was renamed
> > > to Y in 2014"
> > Documentation lives on way longer than expected. People still read and
> > order reprints of the Erlang book published in 1994 (and 1996 for the
> > second edition), some of which are translations.
> > Many older versions of books are what is in libraries and whatnot, since
> > Joe's first version in early 2000s. For people using these versions, you
> > end up with inaccurate terminology regarding half the name of the
> > language.
> > It's a matter of documentation, but it's a matter of trying to do it
> > right to reduce the amount of confusion. If people look for "Open
> > Telecom Platform Erlang" it would be sweet to get the new documentation
> > and content.
> > Maybe it's easy, but it's still part of a roadmap.
> > Alternatively, would 'Open Telecom Platform, a framework that is not
> > just about telecoms' going to be more cumbersome in documentation?
> > > 6) History is history. Those investigating the language will get it. It
> > > startedo ut being a telecoms thing and migrated to a general language. No
> > > problem. Live web sites can easily add in blurbs. Old articles will be
> > out
> > > of date but not from the time frame of when they were written. No big
> > deal.
> > > The sky isnt actually falling.
> > I could see that being made as a decision, yes.
> > > 7) Obviously this one is just frothing. The man could update the next
> > > version of his book with more information, cool tricks, whatever and sell
> > > it as a second edition.
> > Yes. I like to insert a bit of non-serious content here and there.
> > > 8) What "traditional SDK" are you referring to? The LISP standard lib?
> > ;-)
> > > Java? C? Ruby? Haskell? Which one is the "normal" one? Normal is defined
> > in
> > > the context of the language, not in the context of another language? In
> > > fact the vast majority of SDKs for java are third party to the JDK itself
> > > anyway.
> > I went from this thread's usage of SDK as a similar point to OTP.
> > Erlang/SDK if you will. If you want to keep it as Erlang/OTP, that can
> > work, but needs to be significantly better than what it is right now to
> > have an actually measurable impact.
> > Otherwise, we're throwing stuff at the wall to see what sticks, with no
> > proof that it actually helped anything.
> > > 9) Trying to crystal ball the future will only give you a headache. The
> > key
> > > is to move from where yo are to a point where progress has been made and
> > > recursively loop on that algorithm, not be paralyzed by "what if .... ?"
> > >
> > Non-serious content here also. Not to be taken seriously, but I wouldn't
> > be surprised if it were to happen.
> > > You may have been doing Erlang for ages and feel quite the man but the
> > > question really boils down to "what would you like for the future of
> > Erlang
> > > to be?" If the answer to that in your mind is "A niche language that I
> > can
> > > call myself a guru at and everyone looks at me quizzically and puts up
> > with
> > > my eccentricity or dare say arrogance." then the current name and trend
> > is
> > > fine. If the answer is, "A powerful general purpose programming language
> > > for developing applications using functional paradigms and widely
> > accepted
> > > as being the solution to the next generation of software problems." Then
> > > marketing is important.
> > Oh I love that one. I want Erlang to be adopted so much I wrote an
> > entire book about it and put it online for free, without advertisement.
> > This has taken over 3 years of my spare time, because I wanted Erlang to
> > be more accessible. I invite you to visit it at
> > http://learnyousomeerlang.com, and maybe buy an ebook or print copy if
> > you feel like it would be nicer to read that way. If you prefer a free
> > electronic copies, there are scripts on github to convert it to the
> > kindle format, and a wget line in the FAQ to download a local copy.
> > I also kept writing multiple blog posts at http://ferd.ca that guide and
> > show more tutorials about Erlang, use cases, and tries to sell it as a
> > language as a whole.
> > The reason I'm answering to your suggestion negatively isn't that I
> > don't want Erlang to succeed, it's that I do not believe that changing
> > the meaning of 'OTP' from 'Open Telecom Platform' to 'Open Technology
> > Platform' will have a noticeable impact.
> > Some people do ask the question 'but I don't want to do telecoms', but
> > in my experience, people's issues are the following, to a much higher
> > degree:
> > - The syntax is unfamiliar (or ugly)
> > - It's difficult to work with single assignment, recursion, immutable
> > algorithms (most of your algorithm books that rely on arrays with O(1)
> > access to work fine are no longer going to be trivial to translate!
> > That's huge!)
> > - The tooling (rebar, relx, etc.) isn't up to par with other languages,
> > even if it keeps getting better.
> > - Lack of IDEs (that was your prime concern when you joined these lists)
> > - Fighting the idea that "it will be hard to hire Erlang developers" to
> > make it enter and stay in the enterprise.
> > All of those criticism, in the years I've been in the Erlang community,
> > have come up time and time again. They've also have come up orders of
> > magnitude more often than OTP as a name, even if it does come up from
> > time to time.
> > I'm sorry I came up as harsh. I do want better adoption for Erlang and
> > took months if not years of my free time working that way. I do not
> > think renaming OTP is worth the effort, but I'll be glad to be proven
> > wrong through bigger adoption if someone steps up and decides to do it.
> > Now if you please, I'll go back to spending my lunch time working on an
> > post-scripted chapter to the LYSE site introducing maps to people.
> > Regards,
> > Fred.
> erlang-questions mailing list
More information about the erlang-questions