"In other words, Erlang was almost killed by the success of open-source languages."
Re-reading Armstrong's paragraph now, in 2023, I can see how you might think that, but it's not really what happened. Remember, this was 1998. The point about "proprietary language" in the Ericsson Radio decision was partially that you couldn't get a compiler from a competing vendor if you didn't like the one you had, and partially that you had to teach your developers Erlang yourself, since it wasn't widely used outside Ericsson. Open source wasn't big in Ericsson in 1998; typical tools were Solaris, Sun's C++ compiler, Clearcase version control, Sybase and Corba, or Microsoft's C++ compiler and NT.
(Source: I worked at Ericsson Radio in 1997, on a C++ project, and in 1998 and 1999 on Erlang and C-related things, including a year at Ericsson's Computer Science laboratory.)
Thanks so much for commenting and that's a really good point! I've conflated more 'open' languages - in the sense of there being standards and being able to buy compilers from different vendors - with open source as we know it today.
If I've got my dates right then the Open Source Initiative only started in 1998 (although of course the idea of free software has been around for much longer.
I'll find a way to correct this so that it doesn't create a misleading impression. Thanks again.
As you worked on Erlang at Ericsson in the late 1990s can you say anything about how the decision to move away from Erlang was received?
Erlang! Delighted you chose this. I knew it first as a measure of network capacity and then later learned about the language. Thank you for filling in the history.
I've been fascinated by Erlang ever since picking up Joe Armstrong's book about a decade ago. Not just the language itself but also the fact that it's survived when so many other and apparently better supported languages have fallen by the wayside.
Also really like Elixir - just wish I had time to do more than tinker with the language!
In 2007 I wrote much of the business plan for a wireless infrastructure startup. I would have been the non-engineering co-founder, working with two technical co-founders. It was then that I learned about Erlangs as measure of network capacity and also the language.
Based on this topic choice, I’m wondering if you will look at HTTP alternatives (eg MQTT) as potential topics?
That's a really interesting suggestion on MQTT. It would be a natural follow on from Erlang and looking at how standards like MQTT and others emerge and why they succeed or fail feels like a good fit. Leave it with me!
Among Erlang developers, the Erlang ban was initially devastating. But the Erlang projects were relatively few and small, so your average developer in Ericsson wouldn't really have heard of Erlang and saw the ban as a just a reasonable decision to use standard tools (C++, UML, CORBA) instead of messing around with weird stuff cooked up in lab.
The ban almost certainly catalysed the decision to open-source Erlang, and that was a clear win for everyone involved with Erlang.
Probably the most dramatic effect of the ban was that a bit over half of Ericsson's computer science laboratory, including Joe Armstrong, quit and founded Erlang-driven 'Bluetail', which became an overnight success in the dotcom bubble.
That's really interesting - thanks so much. I guess the timing was really helpful being at the start of the Open Source Initiative and the dot-com boom.
"In other words, Erlang was almost killed by the success of open-source languages."
Re-reading Armstrong's paragraph now, in 2023, I can see how you might think that, but it's not really what happened. Remember, this was 1998. The point about "proprietary language" in the Ericsson Radio decision was partially that you couldn't get a compiler from a competing vendor if you didn't like the one you had, and partially that you had to teach your developers Erlang yourself, since it wasn't widely used outside Ericsson. Open source wasn't big in Ericsson in 1998; typical tools were Solaris, Sun's C++ compiler, Clearcase version control, Sybase and Corba, or Microsoft's C++ compiler and NT.
(Source: I worked at Ericsson Radio in 1997, on a C++ project, and in 1998 and 1999 on Erlang and C-related things, including a year at Ericsson's Computer Science laboratory.)
Hi Matthias,
Thanks so much for commenting and that's a really good point! I've conflated more 'open' languages - in the sense of there being standards and being able to buy compilers from different vendors - with open source as we know it today.
If I've got my dates right then the Open Source Initiative only started in 1998 (although of course the idea of free software has been around for much longer.
I'll find a way to correct this so that it doesn't create a misleading impression. Thanks again.
As you worked on Erlang at Ericsson in the late 1990s can you say anything about how the decision to move away from Erlang was received?
Probably because I'm a substack novice, my reply to your question ended up at the bottom of the comments instead of as a child to your comment. Sorry.
Couple of mistakes: "Just as Rails on Rails built on Ruby";
"At the time of the acquisition Erlang employed only 35 engineers."
Hi. Doh! Thanks so much for spotting these. Now fixed.
Erlang! Delighted you chose this. I knew it first as a measure of network capacity and then later learned about the language. Thank you for filling in the history.
Hi Jon. Thanks so much, really glad.
I've been fascinated by Erlang ever since picking up Joe Armstrong's book about a decade ago. Not just the language itself but also the fact that it's survived when so many other and apparently better supported languages have fallen by the wayside.
Also really like Elixir - just wish I had time to do more than tinker with the language!
In 2007 I wrote much of the business plan for a wireless infrastructure startup. I would have been the non-engineering co-founder, working with two technical co-founders. It was then that I learned about Erlangs as measure of network capacity and also the language.
Based on this topic choice, I’m wondering if you will look at HTTP alternatives (eg MQTT) as potential topics?
That's a really interesting suggestion on MQTT. It would be a natural follow on from Erlang and looking at how standards like MQTT and others emerge and why they succeed or fail feels like a good fit. Leave it with me!
User 'jabl' on Hacker News had a reasonable take on some of the politics involved in the ban:
https://news.ycombinator.com/item?id=36570765
Among Erlang developers, the Erlang ban was initially devastating. But the Erlang projects were relatively few and small, so your average developer in Ericsson wouldn't really have heard of Erlang and saw the ban as a just a reasonable decision to use standard tools (C++, UML, CORBA) instead of messing around with weird stuff cooked up in lab.
The ban almost certainly catalysed the decision to open-source Erlang, and that was a clear win for everyone involved with Erlang.
Probably the most dramatic effect of the ban was that a bit over half of Ericsson's computer science laboratory, including Joe Armstrong, quit and founded Erlang-driven 'Bluetail', which became an overnight success in the dotcom bubble.
That's really interesting - thanks so much. I guess the timing was really helpful being at the start of the Open Source Initiative and the dot-com boom.