[petsc-dev] including PETSc4py (and Tao4py) inside PETSc repository

[Barry Smith]

  Maybe someday git will have additional features that allow multiple repos to share things like branches, commits, other stuff? so that having all these packages in different repos is possible, but for now it appears having one repo is just so much dang more convenient.


> On Oct 25, 2015, at 3:12 AM, Lisandro Dalcin <dalcinl at gmail.com> wrote:
> 
> On 23 October 2015 at 06:30, Barry Smith <bsmith at mcs.anl.gov> wrote:
>> 
>>> On Oct 22, 2015, at 10:24 PM, Satish Balay <balay at mcs.anl.gov> wrote:
>>> 
>>> As I said - depending on the precondition for how tightly petsc &
>>> petsc4py should be synced - we can choose the appropriate model.
>>> 
>>> [for the metioned precondition - I was suggestion the model].
>>> 
>>> As you say - the initilal phrasing is wrong - so the model is not
>>> suitable.
>>> 
>>> A single repo does make it easy for 'all feature branches' to be in
>>> sync..
>>> 
>>> In that model with-petsc4py=1 would be the default. However anyone
>>> changing the petsc UI would also have to update petsc4py with the
>>> change.. [and Lisandro says - that change could be buggy - and he
>>> would like to review it -before its applied]
>> 
>>   Just like Jed was going to review everything anybody put into PETSc before it got into master. That is just unrealistic and won't last long. Better to just have a good test suite.
>> 
> 
> OK, Barry, I very much understand all your concerns. My opposition to
> incorporate petsc4py within petsc is not strong. I agree maintenance
> and contribution will be much easier, my concern is just about adding
> more and more stuff to the PETSc repo  year over year. If for the case
> of petsc4py you think that is not a big issue, then I have to trust
> you, it is not easy to catch you on the wrong side ;-).
> 
> BTW, I no longer care about keeping petsc4py compatible with previous
> PETSc releases. While such approach is workable for pure C code (like
> in PetIGA), things get complicated when you add Cython to the stack.
> 
> So, let's go ahead and do the merge.
> 
> 
> 
> 
> -- 
> Lisandro Dalcin
> ============
> Research Scientist
> Computer, Electrical and Mathematical Sciences & Engineering (CEMSE)
> Numerical Porous Media Center (NumPor)
> King Abdullah University of Science and Technology (KAUST)
> http://numpor.kaust.edu.sa/
> 
> 4700 King Abdullah University of Science and Technology
> al-Khawarizmi Bldg (Bldg 1), Office # 4332
> Thuwal 23955-6900, Kingdom of Saudi Arabia
> http://www.kaust.edu.sa
> 
> Office Phone: +966 12 808-0459