Glad to know you are up to speed ...
Try this in v0.4 and in v0.5-dev: length( methods(Base.call) ) v0.4 gives
1000, v0.5 gives 1 (deprecated)
So, Julia is known to work with 1000-fold signatured methods. The fact
that v0.5 is better groomed suggests its worthwhile to consider approaching
your same outcome more parsimoniously or with greater subspecificity ..
lifting some largely clustered highly overlapping signature-driven doings
and letting ... all the oscilloscopes be so scoped, all the multimeters be
multiply metered ... so what you think about as a single method with
signatures everywhichway, is realized as subspecializations of a
[partially] shared abstractions.
abstract Device
abstract Scope <: Device
abstract Meter <: Device
@enum vendor Tektronix, Fluke
type LogicAnalyzer{vendor} <: Scope
...
end
type Multimeter{vendor} <: Meter
...
end
TLA7000 = LogicAnalyzer{Tektronix}( ... )
Hydra2638A = Mutimeter{Fluke}( ... )
handshake{T<:Device}(anyDevice::T) = ...
read{T<:Scope}(scope::T) = ..
read{T<:Scope}(scope::T{Tektronix}) = ..
read{T<:Meter}(meter::T) = ..
On Sunday, March 27, 2016 at 6:03:11 PM UTC-4, Oliver Schulz wrote:
>
> Hi Jeffrey,
>
> On Sunday, March 27, 2016 at 3:51:18 PM UTC+2, Jeffrey Sarnoff wrote:
>>
>> Assuming your handshaking and signal management is keeping information
>> available in some easy to discern and decode manner and you are not pushing
>> latency issues, Julia should be a good platform for that.
>>
>
> I have no doubt of that. :-) Both in terms of execution speed, and other
> features (native coroutines, etc.) Julia is an excellent platform for this.
> I used to do this in Scala (using Akka actors), which was fine, but I
> believe it can be done even better in Julia (at least given certain
> requirements and boundary conditions).
>
>
>> Compilation time usually is not a concern -- and you can make your
>> package autoprecompile [...] pay attention to the general guidelines here
>> performance
>> tips <http://docs.julialang.org/en/release-0.4/manual/performance-tips/>
>> [...] and use the Devectorize package
>> <https://github.com/lindahua/Devectorize.jl>
>>
>
> Yes, I will certainly use precompilation as far as possible (well, not for
> actual user measurements scripts, etc. but for the device implementations),
> and I am familiar with the Julia performance tips and memory management
> recommendations. My question rather revolved around whether Julia's
> dispatch will be happy with O(1000) (or more) methods for a functions.
>
>
