[erlang-questions] Erlang Extension Proposals (EEPs)

Robert Virding robert.virding@REDACTED
Mon Jan 29 22:53:39 CET 2007


Thomas Lindgren wrote:
> --- Per Gustafsson <per.gustafsson@REDACTED> wrote:
> 
>>In conjunction with the Erlang User Conference 2006
>>there was a workshop 
>>concerning the future development of Erlang.
>>
>>In this workshop a decision was made to adopt the
>>Python process for 
>>future Erlang development. Since nothing has
>>happened since then I have 
>>taken it upon myself to edit PEP 1, the python
>>extension proposal which 
>>describes how the python process works. This should
>>be viewed as a draft 
>>so please feel free to suggest changes. There are
>>also a lot of decision 
>>that needs to be made and issues that need to be
>>solved. I'll list some 
>>of them below.
> 
> I think the basic concept sounds useful, since it may
> make proposals more concrete and discussions better
> focused, and it preserves a history of the decisions
> made. Of course, there still may be devils in the
> details.

There are always devils in the details! :-) But I agree with you.

> My first question is: is this intended for changes to
> Erlang, or to Erlang/OTP? Given that the ultimate
> acceptance decision seems to taken by OTP, it seems
> like we're really talking about Erlang/OTP?
> (Alternatively, that the two are basically the same.
> This issue is a bit unclear at present.)

I think that that will, in most cases, be relatively clear. If a change 
is about the language or a small set of core modules then it is Erlang, 
otherwise it is OTP. This assumes that the OTP version of the language 
Erlang also defines the language.

The problem with this is of course that if someone wanted to do their 
own implementation of Erlang there is no real definition of the language 
and EEP will not help here. Another problem is all the stuff in the 
emulator which mightn't be considered part of the language but which are 
necessary to make OTP run. For example boot/startup procedure and some 
ETS tables whichare predefined by the system.

One solution would be to revive the erlang spec and run it in 
conjunction with, and separately from, the OTP emulator. If there is an 
interest.

> And, second question, will OTP henceforth primarily
> use this process to introduce changes to the language?

Hopefully, if the suggestion is adopted, they will. That would also be a 
good way for them to announce their future plans.

Robert



More information about the erlang-questions mailing list