Talk:Four bit adder: Difference between revisions
→I made a latch!: new section
(→C++ code size: Not appropriate.) |
(→I made a latch!: new section) |
||
| (2 intermediate revisions by 2 users not shown) | |||
Line 256:
::I suspect the C++ (not the C) code, rather than being idiomatic, is "enterprise-y" the kind of code you get when people refuse to say no or the people in charge think "I might just need this" or "if I put this in, I can't be fired for leaving it out". There are several examples already existing that could be used to show what is expected from an answer. RC rarely requires any example of that size. It is just as unreasonable to have the C++ example as it would be to add a code golf solution. Simulations are supposed to reflect an ''appropriate'' abstraction of what is simulated. --[[User:Paddy3118|Paddy3118]] 10:26, 1 October 2011 (UTC)
::: Actually, the C++ code looks very idiomatic to me. Certainly not the most efficient, but it's very clear and readable in function. (Actually, frightfully so; I've never seen C++ code that clear) Also, an opinion on the code golf reference...Code golf is generally undesirable because it reduces clarity of code, and adds perverse incentives for example writers. --[[User:Short Circuit|Michael Mol]] 11:42, 1 October 2011 (UTC)
Could someone move the C++ code to a separate sub-page at least? That huge wad of code is over 16 pages long where the others are less than two pages! It is clearly [http://oxforddictionaries.com/definition/anomalous anomalous] in size.
(P.S. it would be instructive to know a little about how the code was written. Was it written for this RC task alone? Was a lot of the code generated by an IDE? ...) --[[User:Paddy3118|Paddy3118]] 04:59, 1 October 2011 (UTC)
: Concur on subpage, and done. I think it was very probably written for Rosetta Code; the usage of a namespace named 'rosetta' is good indicator. --[[User:Short Circuit|Michael Mol]] 11:42, 1 October 2011 (UTC)
== Clojure solution ==
I added a second Clojure solution to show what I think is more idiomatic Clojure code. Because Clojure does not have TCO (which unavailable in Java), it's generally not a good idea to write recursive code in the old Lisp style as done in the original solution ('''n-bit-adder''' function). If possible, you should use the special form '''recur''' in a tail-call position to avoid growing the stack. But most of the time, it's better to find a solution using '''reduce''' if you can.
The second solution shows a couple of interesting language features. First, it uses destructuring to simplify extracting multiple values out of a vector (see my '''full-adder''' function). Second, it uses '''reduce''' in '''nbit-adder''' which is a good thing for Clojure programmers to learn. Also, the pre-condition (''':pre''') is a nice way to check for errors. It can be turned off without having to change the source.
The choice of big endian or little endian is arbitrary, but I think big endian is easier to read. That requires a little extra fiddling with how the bits are fed to the reduce function, but also shows off the '''rseq''' function, which is constant-time reverse for vectors.
== I made a latch! ==
It wasn't my original goal. Instead my first step was a concurrent implementation a little more robust than MizardX's. That method probably would be sufficient to build a latch, but I wanted something more resistant to blocking and deadlocking. Anyway,
<pre>
=== RUN Test-4
--- PASS: Test-4 (0.01 seconds)
logic_test.go:17: power up
logic_test.go:20: S: 0 R: 0 Q: 1 NQ: 0
logic_test.go:22: pulling R = 1
logic_test.go:25: S: 0 R: 1 Q: 0 NQ: 1
logic_test.go:27: pulling R = 0
logic_test.go:30: S: 0 R: 0 Q: 0 NQ: 1
logic_test.go:32: pulling S = 1
logic_test.go:35: S: 1 R: 0 Q: 1 NQ: 0
logic_test.go:37: pulling S = 0
logic_test.go:40: S: 0 R: 0 Q: 1 NQ: 0
</pre>
The concurrency isn't very exotic, and could probably be coded pretty easily in most languages with any kind of concurrency support, but I'm not sure it's quite ready for another task. It still runs asyncronously, with no concept of simulation time, so you have to do the hokey thing of just sleeping for a bit to let it run. I have a vague idea of using a priority queue for simulation time. Maybe that will be next. —[[User:Sonia|Sonia]] 01:05, 9 July 2012 (UTC)
| |||