A little over a month ago Jeff put out a "yt user survey" which had pretty good turnout, along with some very useful and helpful responses. I'm writing with a brief summary (please feel free to chime in, if you have further comments you'd like to bring up here on the list) and a bit of a 'state of stuff' as well, to summarize how this feedback has been received and acted upon.
The breakdown of responses:
This spurned a number of interesting discussions between developers, along with some concrete responses to the feedback. I've summarized these below.
We've registered an IRC channel on irc.freenode.net. You can access this with any IRC client, such as Adium, irssi, XChat [Aqua], and so on. We have also created a web interface to IRC: http://yt.enzotools.org/irc.html . Typically there are three or four people in the channel and a bot that echoes commits to the main yt repo. So far I think it's been a success; please feel free to come visit or ask questions there.
A few developers, led by Cameron Hummels, Jeff Oishi and Britton Smith (with help from me) have written a web GUI. This will be a focal point of the yt 2.2 release, which we anticipate happening in the relatively near future. If you are interested in experimenting with these BETA sequences, feel free to drop by IRC or shoot an email to yt-dev and we can help you get started. Once the final release is out we'll have some documentation that describes how to use it.
The next major build of the documentation will address the primary complaint in the user survey: search in the docs is simply broken. Too many results that are not germane are returned, as a result of the inclusion of many redundant docstring pages. While the API docs are nice, it is clear they overwhelm the search functionality and need to be excluded. Additionally, the recipe system is going to be expanded. Several new recipes have been added, but unfortunately we are now reaching a scaling issue. If anyone has any suggestions how to scale to dozens of recipes/snippets, they would be greatly appreciated!
Multi-step analysis routines are tricky, but Jeff Oishi and I are going to try to include a few examples, including scripts that produce publication plots and analysis. If anyone has any scripts they would like to share, in particular that perform multi-step analysis, they would be greatly appreciated.
We're moving all the old bugs and wiki pages to BitBucket soon, and I'm going to add the ability to file bug reports from the command line. This should enable much easier tracking of items and making sure problems and issues don't get lost. We're also optimistic that BitBucket will help to make developing easier; we've seen this so far with the forking capabilities, and hopefully we'll continue seeing this with the bug tracking and wiki transitions.
If anyone has any other suggestions or feedback, please consider this an opportunity to raise your voice! If you have any ideas in particular about future directions, or feelings on complexity or the documentation, we'd love to hear them. If there's any way you'd like to contribute -- either through developing, or non-technical contributions like documentation, designing how the yt API works, touching up the website or anything else -- please feel encouraged to head on over to the yt-dev list or IRC and strike up a conversation.
Thanks everyone who responded to the survey!