Well, I wrote the code, so I suppose I should comment on it a bit. The
ioc_macros use a very loosely defined version of the state-monad. I don't
know if the functions/macros follow all the monad laws, and I really don't
care, but here's the problem:
I needed to be able to track some state while parsing the clojure code. I
needed to be able to define blocks while defining other blocks, to have
instructions reference each other, etc. All this could be done with a atom,
but I really don't like mutable state. So I decided to write the entire
thing in a fully immutable way.
Thus, the first step is to start defining functions that return functions
that will take a state:
(defn assoc-plan [k v]
(fn [state]
[nil (assoc state k v)]))
So each "monadic function" is a function that takes a state, and returns a
vector of [value new-state]. This allows me to write functions like
add-instruction that both add an instruction to a block and also return a
id to that instruction.
All the other functions in the code are built off this simple concept.
gen-plan then wires them all up together, and is basically monadic "bind",
except it's optimized to use lets instead of nested fns.
All this allows the parsing code to remain rather simple, while tracking
state, and remaining fully immutable. The code below looks as if it is
mutating state, but is actually fully immutable.
(defmethod sexpr-to-ssa 'if
[[_ test then else]]
(gen-plan
[test-id (item-to-ssa test)
then-blk (add-block)
else-blk (add-block)
final-blk (add-block)
_ (add-instruction (->CondBr test-id then-blk else-blk))
_ (set-block then-blk)
then-id (item-to-ssa then)
_ (if (not= then-id ::terminated)
(gen-plan
[_ (add-instruction (->Jmp then-id final-blk))]
then-id)
(no-op))
_ (set-block else-blk)
else-id (item-to-ssa else)
_ (if (not= else-id ::terminated)
(gen-plan
[_ (add-instruction (->Jmp else-id final-blk))]
then-id)
(no-op))
_ (set-block final-blk)
val-id (add-instruction (->Const ::value))]
val-id))
Most of the time, I recommend people start reading the at functions such as
add-block, get-block, set-block, etc. Ignore the implementation of
gen-plan, and just examine how it's used. After about 30 minutes the light
will go on, and it'll all make sense.
Pretty printing (via the debug function) the output of
parse-to-state-machine will also help you make sense of what's being
constructed.
I hope this helps.
Timothy Baldridge
On Fri, Jul 26, 2013 at 8:51 AM, john <john.vie...@gmail.com> wrote:
> Hi,
> I am trying to understand the code in
> ioc_macros.clj<https://github.com/clojure/core.async/blob/master/src/main/clojure/clojure/core/async/impl/ioc_macros.clj>
> .
> I stumbled on the macro gen-plan which doc reads "Allows a user to define
> a state monad binding plan"
> I am wondering what makes the macro a "state monad" and how does it
> simplify the building of the state-machine data structure?
>
> Many Greetings
> John
>
>
>
>
>
> --
> --
> You received this message because you are subscribed to the Google
> Groups "Clojure" group.
> To post to this group, send email to clojure@googlegroups.com
> Note that posts from new members are moderated - please be patient with
> your first post.
> To unsubscribe from this group, send email to
> clojure+unsubscr...@googlegroups.com
> For more options, visit this group at
> http://groups.google.com/group/clojure?hl=en
> ---
> You received this message because you are subscribed to the Google Groups
> "Clojure" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to clojure+unsubscr...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>
>
--
“One of the main causes of the fall of the Roman Empire was that–lacking
zero–they had no way to indicate successful termination of their C
programs.”
(Robert Firth)
--
--
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts from new members are moderated - please be patient with your
first post.
To unsubscribe from this group, send email to
clojure+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
---
You received this message because you are subscribed to the Google Groups
"Clojure" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to clojure+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
