Suggestions on project developments

[S Armour]

I am an EE/Optics Student and have been working on developing some tools to streamline the workflow with some of the NumFOCUS affiliated libraries. The code that I have written is in the nature of " I got to get this done and turned in"  but it is stuff that as a working scientist is defiantly needed. So how do I solicit this to the library developers to then possibly get it integrated into the main library? For example, NumFOCUS is involved with YT and xarray but since xarray is pandas in greater than 2D and pandas has calls for plot making from its dataframe's with matplotlib then it seems that xarray should have built-in calls to create YT "graphs". And  YT should add a loader for xarray since its a lot easier to work with xarray then some HD5 data structure that is more of an astro thing vs xarrays is a pure python extension to pandas? Wich would then make YT the mainstream way of visualizing volumetric data in the python ecosystem. Wich we need since vispy has gone nowhere and YT better than that VTK wrapper that is the current standard.  

[Nathan Goldbaum]

Hi,

I'm a yt developer. For your specific example I definitely agree with you that closer integration between yt and other NumFOCUS projects would be great. We have several project ideas related to this, including an ongoing effort to make the code less astronomy specific. We also recently moved to git/github from hg/bitbucket and we're hoping that going forward there will be fewer barriers to entry for interested contributors.

As in all things open source, it takes someone to put the time in and implement whatever vision you have. The best way to ensure any given project gets done is to contribute it yourself or indirectly encourage a contribution e.g. by funding it or finding a student to work on it. At a minimum you could start a conversation going in the communities that you would like to influence.

I'd also like to say that vispy, mayavi, yt, and non-python applications and libraries like VTK, visit and paraview all fulfill different niches in the 3D data viz space. For example, vispy is more focused on GPU-powered 3D vis while yt is more focused on providing an API for interactive and scripted data analysis in Python. I'm glad you think that yt might be useful for your workflows, but I also don't want to make value judgements about other libraries.

Nathan

On Sat, May 6, 2017 at 6:38 PM S Armour <strin...@gmail.com> wrote:

I am an EE/Optics Student and have been working on developing some tools to streamline the workflow with some of the NumFOCUS affiliated libraries. The code that I have written is in the nature of " I got to get this done and turned in"  but it is stuff that as a working scientist is defiantly needed. So how do I solicit this to the library developers to then possibly get it integrated into the main library? For example, NumFOCUS is involved with YT and xarray but since xarray is pandas in greater than 2D and pandas has calls for plot making from its dataframe's with matplotlib then it seems that xarray should have built-in calls to create YT "graphs". And  YT should add a loader for xarray since its a lot easier to work with xarray then some HD5 data structure that is more of an astro thing vs xarrays is a pure python extension to pandas? Wich would then make YT the mainstream way of visualizing volumetric data in the python ecosystem. Wich we need since vispy has gone nowhere and YT better than that VTK wrapper that is the current standard.  

--
You received this message because you are subscribed to the Google Groups "NumFOCUS" group.
To unsubscribe from this group and stop receiving emails from it, send an email to numfocus+u...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.