The section below shows a representative sampling of free-form comments left by the survey participants.
On switching from IDL to Python:
I recently switched from using IDL as my primary programming language to using Python. (Graduate student)
Mainly IDL user who wants to switch to Python as it is more open source. (Graduate student)
I learned IDL as an undergrad and continue to mostly code in it as a grad student. However I’ve been learning Python lately and plan to mostly switch over within the next year or so. (Graduate student)
I plan to learn Python but haven’t yet worked with it. (Graduate student)
At this stage I see Python as the future and am rapidly moving away from IDL. (Graduate student)
I learned IDL as an undergrad (class of 2004) and used it nearly exclusively [...] until about two years ago. Over the last two years I’ve been slowly switching to Python [...] (Postdoc)
I’ve only recently started working in IDL and Python. I expect to do quite a bit of development in Python from now on. (Postdoc)
I want to learn Python and R as soon as possible (Postdoc)
While I haven’t learned it yet many of my colleagues use Python and I make all of my students learn that (instead of e.g. Matlab). (Faculty)
I am telling all of my students to learn Python and through that I am also gaining proficiency in Python. This is different from what my advisor did. He told me to use the language that he used so that he could help me debug it. (Faculty)
Moving to Python as IDL needs a license... And I just like the language. (Faculty)
Currently I code in IDL but I am trying to switch to Python. (Faculty)
I plan to switch from IDL to Python over the next 2-3 years. (Faculty)
I’m intending to try out Python soon. (Faculty)
Taking a workshop on Python soon I can encourage and help my students learn a language that can be used outside of academia as well. (Faculty)
On the desire for more opportunities to improve software development skills:
If there is any software development training program for astronomers I would love to attend. (Graduate student)
I think it should be strongly recommended that people going into astronomy should take programming classes. In fact I would make it part of the required course work to get a B.S. in Physics or Astronomy. Most astronomers I know did not take any formal programming course we learned as we went along. Most of us write our own code and many do not use good coding practices making reading or adapting code from other astronomers a lot more painful than it should be. (Graduate student)
Helps to include courses in computation and statistics in grad curriculum. (Graduate student)
Astro students should get more formal training in programming and software design (and it should happen at the undergraduate level whenever possible). (Graduate student)
I wish I could get more formal training on programming. I have the feeling that Astronomy and Physics department usually don’t emphasize the importance of programming until people start doing researches. (Graduate student)
The coding skills incoming graduate students possess seem to vary wildly but they are often dismal with respect to the level required to begin doing serious research right away. In my department there seems to be little motivation to rectify this with either: (1) requiring undergraduate CS preparation as a condition for admission; or (2) organizing a programming course for beginning graduate students. This situation seems if not unsustainable very much non-optimal for the cultivation of strong substantive independent research skills. I imagine many other departments are currently facing the same dilemma. (Graduate student)
More formal training in software use/development would be wonderful. (Graduate student)
Formal training on astronomical software packages as part of my astronomy undergraduate degree would have been very helpful for me. (Graduate student)
We need to be teaching undergraduates and graduates good object-oriented design skills from day one. Software is more important now than it ever was and the “learn as you go” mentality causes a tremendous amount of wasted efforts as bad code has to be rewritten all too often with the side effect of having astronomers with career skills that aren’t as well developed as they could have been should they choose to leave astronomy. Also we should make a concerted effort to rewrite some of our legacy tools (IDL libraries, IRAF, etc.) in a language and style that is more easily and cleanly extended and maintained. Incentives to put in the time and money for these [initially] low impact projects are hard to come by though. (Graduate student)
On the lack of career opportunities for people who write software:
People who develop software that the community use should be recognized more for their efforts! (Research Scientist)
It would be great to have more career options for researchers who focus on software development in the astrophysical community. Too many good researchers who contribute lots of great software to the community have been forced out of the field because of lack of recognition for their work and lack of funding for people other than those who publish several science papers a year. (Postdoc)
Software development is evidently as important a tool in modern science as mathematics and just as it has historically not been deemed wise to outsource all mathematics to professional mathematicians I believe a large fraction of scientific software development will have to be accomplished by scientists who are intimately familiar with the problem at hand. Perhaps more than is the case for mathematics though the paper metric used for hiring scientists often pushes excellent software developers out of science and into industry who are then lost to us. (Postdoc)
I find that my observational colleagues are often unaware that we as computational scientists need to write proposals for supercomputers like they write for telescopes. Also when we write science proposals (e.g., NASA, NSF) we have to lie about how much time we will spend developing code say only a few months when in reality it occupies most of the grant period since code development is frowned upon (except within the new NASA ROSES program PDART started in 2014). (Research Scientist)