The Year in Emacs
I have spent the last 3 or 4 years slowly getting a reputation for
extending Emacs in mad ways. So much so that fsbot on #emacs will tell
most problems can be fixed by writing more elisp
I will be continuing this trend in 2013. There's a lot going on so I thought I'd write a summary of everything that I see happening.
Elnode, the EmacsLisp async webserver, will have a v1.0 release some time this year. Elnode is very stable there are just a few things missing, asynchronous logging being one.
A release will fix the Elnode API but we'll start working on version 2 straight away, possibly integrating websockets into Elnode.
Things that will potentially go into Elnode v2 are:
- a proxy
- built in, actor like concurrency
- stuff to make it easier and quicker to write apps, like:
- cleaner APIs
- Emacs modes for investigating logs and such
- Emacs modes for controlling lots of Elnode instances
- I might deprecate the wiki in favour of Elwiki
EmacsWiki reboot with ElWiki
This is a project I am only peripherally involved in but I hope to step up my involvement this year and help get it live. After all, it was a key reason I wrote Elnode.
Achieving this could be a massive win for Emacs. Elwiki sits on top
of creole which is all
written in EmacsLisp and designed to be extended and hacked in an
Emacs-y way. Elwiki pages will be able to contain lisp, which I see as
a big win for doing sophisticated things in the transformation (I hope
we can do things like pull in bits of other pages or even SQL
querys). Elwiki pages can also contain
Org mode tables and
formulas, so Elwiki pages could be spreadsheets.
Because we have a Lisp wiki engine that will run in Emacs I also see us being able to deliver a fully federated EmacsWiki with it. You'll be able to install the EmacsWiki package and select another EmacsWiki node and the node's git repo will be cloned... we'll have more code, possibly based on magit to monitor the changes between repos so you can track where pages in your local copy are differing.
Elwiki will also have a native creole editing mode for Emacs, so you'll be able to edit in simple text in an Emacs buffer.
I think federation could help achieve scalability for the main EmacsWiki because we could distribute rendering, potentially to any node hosting the federation. The Elwiki is also being written to be more cachable.
I hope we can do all this without anyone doing any transform work on the page content itself. I'm pretty sure we can just pick the content up from Git and provide it on Elwiki and where we can't I'll hack creole until we can. We'll do that in a stages of course, as soon as it's ready I'm hoping we'll be able to persuade Kensanata to bless the process.
Marmalade in Elisp
Marmalade is written in node.js and I have to run it in a virtual machine which is expensive in terms of computing resourcces (and therefore money).
I did quite a bit of work last year on getting to a state where the Marmalade mongo db could be pulled apart. I was hoping to base the new elisp marmalade on the old mongo db database but now I think I will just convert it to something much lighter weight, files on a disc for example. It should be easier to get the job done that way.
Elpakit includes a lot of the work needed to make a new marmalade so I feel like I'm nearlly there on this one.
I feel like there's a ton of extra stuff marmalade could help with:
- signed packages
- secured encrypted packages (for storing your config in)
- personal application repositories
- federated package archives
- like mirroring but able to push and pull content around
- integration with melpa?
Most of these are related to elpakit in some way.
Caroline and I will be continuing with TeamChat.net apace. I hope we'll be able to open the doors properly at the end of January. Adding database scaling has been much harder to do over December than I thought (although now it will be easy for other people).
Erwin is now in a position where his brain can expand with the addition of webservices. We might even get to the point where Erwin web services can be offered from some sort of directory to any teamchat instance. That would promote sharing of stuff like github alerters.
I'll be working throughout the year to take parts of TeamChat, particularly shoes-off (the IRC bouncer) into a state where they can be used independently.
Elpad is an HTTP based async editing environment. It's going to allow Emacs users to share editing text, live, with web users or other Emacs users. It will be way easier than the faff around rudel; there will be a simple client and the ability to switch a buffer into live editing mode with a particular person.
I'm really excited about Elpad, it's another one of the core reasons I wrote Elnode. It's so annoying that, in my editor I have to put so much effort into sharing code with someone else. It should be trivial and dynamic and with Elpad it will be.
Elpad will also be integrated into TeamChat and Emacs users will be able to naturally select bits of text and move them into an Elpad environment with someone from their teamchat irc.
Elpad has the potential to be a disruptive technology. It could be used to promote pairing and sharing code better around the world. I really hope the people on #emacs will take to it.
2012 had a few attempts at multi-processing but in 2013 we'll get this right I think.
John Wiegley and I, independently came up with the same idea, to spawn a child Emacs to do some work and communicate results back to the main Emacs. John and I's approach differed only in the intent, his was to get a functional result back and mine was to get a stream of output back.
I think it's obvious these should be combined to give the best of both worlds: the child Emacs should deliver a stream of results. It looks to me like some sort of actor framework is needed.
What was interesting is that both our approaches suffered from the same problem, how to create the right contextual environment in the child emacs. That's a problem that remains to be solved. My current thinking is that a number of different approaches are needed so user's can select which is most appropriate to them.
Excitingly Tom Tromey has been working on adding threads to Emacs for a while and now has a branch that more or less works. Emacs 25 might even have threading added.
Skiiny is my blog engine and I'm going to be working on that this year to add RSS (a glaring omission from the current code) as well as a ton more configurability and extensibility.
Maybe I will even offer some Skinny hosting as a buy service this year.
I'll definitely get Caroline to stand around in her Skinny jeans so I can have a picture of her legs as a logo.
Maildir is my Elisp mail user agent. It's a completely personal tool right now, I don't think anyone else uses it at all. But I'll be working this year to add foldering, indexing, moving the heavy processing things it does into multi-processing and making an Elnode app to sit on the MTA for the purpose of sucking the mail out (I currently use rsync for that).
Frankly, I don't care if anyone else wants to use this one, this one's just for me.
Elpakit is less than a few months old but already is very embedded in all my workflow. I'll be working to make it more and more useful through the year. Specific things I see are:
- building info and HTML manuals from texinfo source and perhaps from creole
- making it trivial to run travis builds
- perhaps even detecting certain standard elisp test frameworks? ecukes?
- letting package lists be stored in files or at urls, in json perhaps
- using magit in some way to monitor repo state of elpakits
- being able to clone all the repos from an elpakit to another location
Other Emacs things
Hopefully at some time this year someone (maybe me, but probably only
if I can make Emacs my day job) will fix the
select problem in
devpoll for platforms that
have it seems the right thing to do.