Re: Build agenda asynchronously

[Tim Cross]

"Dr. Arne Babenhauserheide" <arne_bab@web.de> writes:

> [[PGP Signed Part:Undecided]]
>
> Tim Cross <theophilusx@gmail.com> writes:
>
>> Emacs' support for asynchronous operations is at best primitive. There
>> is built-in support for calling processes asynchonously and
>> there is some other development work to set the stage for adding threads,
>> but I think general asynchronous processing inside Emacs is a long way
>> off. A lot of how Emacs lisp works fundamentally lacks the low level
>> control structures necessary to make data structures and operations on
>> those structures thread safe. This means you have to work at a very low
>> level in order to ensure code is thread safe and that simply isn't
>> practical. Even defining the basic model for an asynchronous emacs lisp
>> is non-trivial and once you have the model, you ahve to implement it.
>
> Maybe it could be possible to fire up a second Emacs and retrieve the
> agenda-buffer?
>

Yes, I've seen people who have done things like that. They have their
'main' emacs instance running where they 'do stuff' and another instance
running where they read mail, check their agenda/calendar or run other
things which tend to 'block'.

There are some issues with this approach though. Some of which are not
obvious.

- Emacs can consume a fair amount of memory. Running multiple instances
  can use a fair amount of memory.
- Startup times make it less appealing as a 'run it up when you need it'
  solution.
- You need to have an understanding of the interaction with various
  cache and other disk files Emacs uses to ensure each instance doesn't
  tread on the toes of the other. 
- A 'strategy' with regards to reverting buffers to file contents.
  Consider adding a todo in your agenda while at the same time, working
  on a related org file in the other instance.  

Provided you know the rules, can operate in an understood disciplined
manner and have a system with sufficient resources, it certainly can be
a viable approach.

Personally, I took a different route. I keep the number of files which
contribute to my agenda to a minimum and have an easy way to
update/change that list. I can quickly switch agenda contexts depending
on what I'm doing. For example, when I'm at work, I'm not interested in
any of my 'home' tasks, events, etc. As a result, generation of various
agenda views is fast and I don't need to wait. I do have an 'everything'
context, so when I do need that, I can get a complete overview. It does
take longer to generate and will block, but I rarely need it. My
calendar.org file, which contains all my meetings is shared across all
views so that I avoid 'double booking'. I do tend to put all meetings in
Google calendar because it makes it easy to share them with other
devices (tablet, phone, laptop, other people), but I use my icsorg
script to import that into my calendar.org file.