rearranged documentation #129
Conversation
|
We need a way to view the result before accepting/merging, not only the sources. Any idea? |
shssf
added
[WIP] Work in progress
Documentation
|
@fschlimb I am not sure how to do that. Maybe downloading these specific changes from my branch can help. But I will try to find another way and will keep you posted. |
docs/source/GettingStarted.rst
Outdated
|
|
||
| conda create -n HPAT -c ehsantn -c numba/label/dev -c anaconda -c conda-forge hpat | ||
|
|
||
| Building from Source on Linux |
There was a problem hiding this comment.
We have to avoid duplication of build instructions. Building HPAT from sources should be the exception. Hence I'd prefer this to be in the README on github. The instructions in #142 have been verified.
docs/source/GettingStarted.rst
Outdated
| Automatic Parallelization | ||
| ------------------------- | ||
|
|
||
| HPAT parallelizes programs automatically based on the `map-reduce` parallel |
There was a problem hiding this comment.
HPAT is not only map-reduce. It can also do join, groupby etc which are more than that.
It uses domain-knowledge (pandas-like data analysis) to efficiently map the computation to distributed memory.
docs/source/GettingStarted.rst
Outdated
| File I/O | ||
| -------- | ||
|
|
||
| Currently, HPAT supports I/O for the `HDF5 <http://www.h5py.org/>`_ and |
There was a problem hiding this comment.
we currently do not support hdf5 anymore.
docs/source/GettingStarted.rst
Outdated
| one processor only since all processors have the same copy. | ||
|
|
||
|
|
||
| Strings |
There was a problem hiding this comment.
string support has been 'moved' to numba.
|
|
||
| #autosummary_generate = True | ||
| autosectionlabel_prefix_document = True | ||
| autodoc_mock_imports = ['numba', |
There was a problem hiding this comment.
I am no longer sure if autodoc is the right tool for user documentation. I currently do not see how it can help us providing the information we want to give to the user.
It remains to be seens if autodoc can be used for internaal developer docs.
There was a problem hiding this comment.
autodoc is useful if we have standardized docstrings inside our code to document the functions. In that case we can use it for both users guide and developers guide. Users will have information on how to use the functions and arguments whereas developers will have information on how to change the functions. So in current PR it is just used for distributed.py module as an example but I am working on using it for other modules.
| @@ -22,10 +23,11 @@ regression program using HPAT:: | |||
| w -= np.dot(z, X) | |||
| return w | |||
|
|
|||
| This code runs on cluster and cloud environments using a simple command like | |||
| `mpiexec -n 1024 python logistic_regression.py`. | |||
| This code runs on cluster and cloud environments using a simple command like:: | |||
There was a problem hiding this comment.
dot is not the one use case we want to show in a quick intro. it should be something in pandas.
|
I will soon submit new changes in this PR which has a different arrangement of the content and that is better than the one currently shown in this PR. @fschlimb thank you for your comments and they are very helpful. Once this PR is merged than we all can work together to make changes in the content inside every section. |
|
@rdesai16 Is this PR ready to merge? |
|
@shssf Yes it is |
* rearranged documentation * further changes to rearrange the documentation * rearranging and renaming some sections
This PR adds updated sphinx setup to enable automated documentation