numpy.org
numpy.org copied to clipboard
numpy
User stories - tracking issue
User Story No. 1 Lotte is a library author that depends on NumPy. She is looking for information about major changes and a release date of the next version of NumPy. She would like to easily find it on the website instead of contacting the core team.
User Story No. 2 Yu Yan was introduced to NumPy in her first week of the Foundations of Data Science class. She is looking for a NumPy tutorial for absolute beginners in Mandarin.
User Story No. 3 Tiago is a software developer. By day, he builds enterprise applications for a Fortune 100 company. By night, he cultivates his academic interests in statistics and computer science using various Python libraries. Tiago has an idea for a new NumPy feature and would like to implement it. He is looking for information on how to contact the person(s) in charge of such decisions.
User Story No. 4
Let me try one:)
Maria is a lecture in the new Data Science department. She needs to teach an introductory class about NumPy and numerical computing to students that know basic Python and math. She is looking for material that can be used for in-person teaching in class, rather than tutorials aimed at self-learning.
User Story No. 5
John is a data scientist, consuming libraries that use python like TensorFlow. He is running into a problem in his code that emits some strange error, and the traceback shows an exception from NumPy. What should his next step be? When should he open an issue on the NumPy issue tracker?
User Story No. 6
I am a NumPy and SciPy developer, and although I have been using NumPy for many years, I still have occasion to check the online reference documentation for a function in the latest released version. I would like to be able to get to the API documentation with just one click from the main page (https://numpy.org/).
For a nice example of what I would find useful, take a look at the pandas web page. The links to the reference documentation are right there in a box on the right side. With just one click I can get to the reference docs of the latest released version (or the development version, or older versions, too).
User Story No. 7
I would like to contribute to NumPy more actively. Since NumPy is such a huge beast, wrapping my head around the functionalities of NumPy is really challenging. I think having a high-level overview / descriptions of NumPy and each of the module would be very helpful. Also, it would be great if C API documentation could have more structure, maybe something to python API. I found it hard to browse through the C APIs.
Adding to @WarrenWeckesser comment, I also like the way scikit-learn webpage separate out the API documentation from tutorial and have links from API to tutorials and examples. Maybe it's just because I'm more familiar with their docs.
User Story No. 8
Kartoffel Schmidt is a data scientist/researcher who's writing code for NumPy, and he knows that at some point, he may have to translate his code to use distributed computing, GPU computing or sparse computing. What guidelines should he follow to make sure that his code will be portable across Dask, CuPy or PyData/Sparse in the future?
User Story No. 9 Christopher writes a monthly column on technology and science in WIRED. His next story is covering the subject of Data Science. He is looking for the interview subjects within the open source Python community who are recognized experts, pioneers, and evangelists for the field.
User Story No. 10 Daphne is preparing a keynote speech for the upcoming conference. She needs a high-resolution copy of the NumPy logo to add to her slides.
User Story No. 11 As a data scientist, Anika needs to stay up-to-date on the latest information about the software packages and libraries she uses at work. Anika would like to receive it in her inbox.
User Story No. 12 Noor used NumPy extensively in her scientific research and would like to acknowledge it in the academic paper she is about to publish. Noor is looking for an example of citing NumPy.
User Story No. 13 Tania is a packager (linux distro, conda-forge, homebrew, containers). What are best practices to build, test, and package numpy? ~Should they bundle OpenBLAS or delegate to the system BLAS?~ Which versions of system dependencies does numpy require? Tania would like some general guidance and links to maintained high-quality packages (conda-forge, debian) to not bother core developers and to not deliver under-performing software to their users.
Packagers definitely need their own story, thanks.
Should they bundle OpenBLAS or delegate to the system BLAS?
Everyone should delegate to the package manager provided BLAS. Only NumPy itself should break that rule, because of PyPI/pip limitations.
User Story No. 12 Noor used NumPy extensively in her scientific research and would like to acknowledge it in the academic paper she is about to publish. Noor is looking for an example of citing NumPy.
Based on #406, placing a link to https://numpy.org/citing-numpy/ in the footer.