Good think I brought this up. Let's get this synchronized. No reason to 
duplicate work. We can discuss the plan on petsc-dev

  Why not just have PetscOptionsInsert() and have it auto-detect the file 
format making life easier for user than remembering half-a-dozen function names?


   Barry


On Mar 14, 2011, at 3:31 PM, Blaise Bourdin wrote:

> Hi,
> 
> No, I had no idea that Sean was bringing an LSU student to work on this.
> 
> I have a sophomore CS student starting right now. The idea is to implement 
> PetscOptionsInsertJSONFile(MPI_Comm comm,const char file[],const char 
> section[],PetscBool  require)
> where every child of section would be parsed an inserted.
> i.e. parsing the following json file with section="petsc"
> {
>  "version": "1.0",
>  "encoding": "UTF-8",
>  "mystuff": {
>      "key1": "val1",
>      "key2": [ "a", "b", "c"]
>  },
>  "petsc": {
>       "sub1": {
>               "key1": "val1",
>               "key2": [ "a", "b", "c"]
>       },
>       "sub2": {
>               "key1": "val1",
>               "key2": [ "a", "b", "c"]
>  },
> }
> 
> would have exactly the same result as using
> -sub1_key1 val1 -sub1_key2 a,b,c -sub2_key1 val1 -sub2_key2 a,b,c
> on the command line.
> 
> My goal is to be able to use a single json file to control an entire 
> computation (mesh generation, preprocessing, analysis, postprocessing, 
> visualization). It seems to me that this is implementable with minor 
> incursions into the petsc code base.
> 
> Blaise
> 
>> 
>> Is this in any related to Sean's plan to have this done this summer with a 
>> student from LSU?
>> 
>>  I could assume, likely correctly, this is one and the same project, I just 
>> want to make sure.
>> 
>>  Barry
>> 
>> On Mar 14, 2011, at 1:31 PM, Matthew Knepley wrote:
>> 
>>> Does anyone have a favorite? Blaise is going to have a student add JSON 
>>> support
>>> to PetscOptions.
>>> 
>>>  Matt
>>> 
>>> -- 
>>> What most experimenters take for granted before they begin their 
>>> experiments is infinitely more interesting than any results to which their 
>>> experiments lead.
>>> -- Norbert Wiener
>> 
> 
> -- 
> Department of Mathematics and Center for Computation & Technology
> Louisiana State University, Baton Rouge, LA 70803, USA
> Tel. +1 (225) 578 1612, Fax  +1 (225) 578 4276 
> http://www.math.lsu.edu/~bourdin
> 
> 
> 
> 
> 
> 
> 


Reply via email to