[petsc-dev] reminder never use #include "mylocalinclude.h" in PETSc source

[Barry Smith]

On May 5, 2010, at 2:21 PM, Jed Brown wrote:

> On Wed, 5 May 2010 13:36:34 -0500, Barry Smith <bsmith at mcs.anl.gov> wrote:
>>   Hmm, didn't know bout this. But it then is not using the xcode
>>   build environment at all (just editor, source code browers
>>   etc). Doubt it will work for building an iphone app :-)
> 
> The behavior of PETSc applications can be modified too much at runtime
> so it falls under the "no interpreters" clause and will be rejected by
> the iPhone store.

   shhh, they don't have to know that.

   But, I think you are wrong, there are apps with many options that are allowed. For example, there is this great graphing app that allows you to enter mathematical formula for equations and it graphs them. It is definitely interpreting the equations.

>  If programs carefully avoid ever calling
> *SetFromOptions() and don't forward the arguments, it will still be
> rejected under the "no libraries" clause.

   Again, they don't have to know this.

   BTW: there is another category of Apps that are used within one institution that don't need to go through the some approval process. I need to investigate this further.

  Also, if we distribute the build process, so long as people are willing to pony-up the 99 dollars to Apple they can install iPETSc themselves directly on their own i's from source. Not that anyone in their right mind would pay anything to run a PETSc example :-)

  BTW: trying to push the "build system" to work for these "broken" systems is a way to improve it; I've already found a couple of minor bugs.

   Barry

> 
> Jed