geda-developers team mailing list archive

[Bob Paddock <bob.paddock@xxxxxxxxx>]

On Tue, Dec 18, 2012 at 8:26 PM, Edward Hennessy <ehennes@xxxxxxxxxxxxx> wrote:

>>   it.  If it makes better sense to use C++ & Qt, that's fine too.  Yes,
>>   the same goes for file formats.

For file formats take a look at YAML.  Lets not invent our own format,
when parser exist of may languages.

> May also want to put C# on the list.

Please no.  It is an absolute nightmare to install a program in any
.NET thing on a thousand different computers that are all configured
differently, ie. Windows.
Spent most of the last year cleaning up the mess left by someone that
used C# without realizing that deployment is not easy, because no two
Windows systems are the same.

> I'm still on the fence with the asynchronous methods.

Look at Herb Sutter articles on his web site and Dr. Dobbs.
What you are after is Active Objects.

> - functions returning an IEnumerable<T> using yield return/break;
> - Parallel.foreach
> - container extensions, such as ConvertAll, Select, Find, OrderBy, etc...

All of which Boost.org  for C++ offers.  Also has Polygon code and
look at the Geometry library.
There a lot of similarities between cartography and routing.
Really it has everything needed but the hairy math specific to layout,
an the GUI.

No one is going to be jumping up and down with joy with the thought of
using C++, however it will get the most common conciseness.

If you want better languages then it is time to move to  languages
like R, Haskell or Erlang.
Some of which are well suited to hairy mathematical problems, like R.

Locks and Threads do not scale.  Any modern system is moving to pure
message passing.  144 core processors ares are already available if
you like Forth.
"But if you have multiple processes sharing and modifying the SAME
memory this is a recipe for disaster -- madness lies here."  -
http://pragprog.com/articles/erlang
Check  that link out for the two paths that were taken Threads/Locks
vs Message Passing...

Better to figure out the hard stuff first, like the modeling, before
worrying about how to implement it.

-- 
http://blog.softwaresafety.net/
http://www.designer-iii.com/
http://www.wearablesmartsensors.com/