Menu
▾
▴
Re: [brlcad-devel] Only Three Weeks Left
|
From: John A. <Da...@co...> - 2008-07-28 23:30:32
|
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=UTF-8" http-equiv="Content-Type">
<title></title>
</head>
<body bgcolor="#ffffff" text="#000000">
Dawn,<br>
<br>
Looks like you have your remaining time well planned. Please keep
me informed of your progress.<br>
<br>
Your question regarding the parser is outside of my technical
expertise. I would defer to Sean's suggestion. Perhaps another
developer reading this might have a suggestion?<br>
<br>
<br>
-John<br>
<br>
<br>
Dawn Thomas wrote:
<blockquote
cite="mid:692...@ma..."
type="cite">
<pre wrap="">Hi John,
I have around 100 hours of working time left till August 11th.
The things on my TODO for completion are as follows in decreasing
order of priority.
1. sorting out exposure of STL elements -
2. solving segfault in the constraint solver -
3. Complete the implicit constraint cycle: 30 hours
3a. Freeze variable expression syntax which also involves
writing the associated functors
3b. Write generator wrapper/ object around BinaryNetwork Constructor
3c. Write Geometry updater object & corresponding function in librt
to update the geometry as a result of constraint solution
3d. Write rt_param functions for at least 3 primitives.
3e. Test above by construction of geometry.
4. Complete the explicit constraint cycle: 20 hours
4a. Freeze Constraint expression syntax /
what forms of constraints to support in the first iteration
4b. Write functors
4c. Test a constraint solution : 1. Addition of constraint
object to the .g file 2. Calling the pc library 3. Parsing of the
constraint
4. Solution display.
5. Add basic mged interfacing: 20 hours
I would love a discussion about what are the expectations
in terms of command interface. Sean can give me some pointers as to
what to add and what not to.
6. Improvement of the solver: 15 hours
These are additions in terms of deterministic algorithms to
reduce the search space and should not take an immense amount of
effort / time now or afterwards.
I have left around 15 hours for the stl issue and the segfault and any
unexpected problems.
</pre>
<blockquote type="cite">
<pre wrap="">We need to decide where a reasonable stopping point is for the SOC.
Keep in mind that there is a week at the end to "scrub code, write tests,
improve documentation, etc.". I think we should plan to finish with a well
documented project that another developer could continue working (even if
you plan to continue working on this after SOC ends).
</pre>
</blockquote>
<pre wrap=""><!---->
Regarding documentation, I think I will concentrate on it during the
11th to 18th period.
I definitely am going to continue working after SoC, but as you said I
understand that it is necessary to have a well documented system which
another developer could also work on.
</pre>
<blockquote type="cite">
<pre wrap=""> I would like you to send me some email about where you think a good
stopping point is, how you will use the remaining time, and what (if any)
additional documentation needs to be prepared.
</pre>
</blockquote>
<pre wrap=""><!---->
So in short I think a good stopping point would be
1. Complete Implicit constraint cycle
2. Explicit constraint cycle as mentioned above
Immediate additional work post gsoc would be
1. Completing the explicit constraint system in terms of geometry
updation ( which should be straightforward except for multiple
solution cases )
2. Implementation of a Hypergraph.
3. Improving the Solver by adding more consistency techniques.
4. Explore symbolic and analytic solution systems.
Sean was suggesting the usage of something less "custom" for parsing
operation. I have used spirit which is I believe resource intensive .
But due to the easy integration with the phoenix system it provides a
good way of utilizing functional programming concepts if necessary.
But I believe the system of having to rewrite grammars in complicated
ways is a drawback. What would be a better solution, using a generic
parser for "most" of the input and using spirit for the specific
expressions parsed from the input by the generic parser? Or use
generic parser for all the operations ?
Sincerely yours,
</pre>
</blockquote>
<br>
</body>
</html>
|