Code Review Software

[Eric Haavind-Berman]

Hi Statalist,

I am worndering if anyone knows of code review software that works with Stata? I am interested in at least having the ability to upload data and code along with some comments about it. Then, when edited, the code shows changes and comments from the reviewer.

I know GitHub is the go-to, but I am curious about what else is out there.

Thanks,
Eric

[Anders Alexandersson]

By "code review software" I assume you mean "version control software" rather than debugging tools such as Stata's trace. GitHub is a web-page for the Git repositories.

Mercurial is perhaps the main alternative to Git. For example, as mentioned in the Stata 15 wishlist, Sergiy Radyakin now prefers Mercurial over Git. Some discussion about differences between Git and Mercurial are here.

[Eric Haavind-Berman]

Hi Anders,

I am actually not looking for version control, I am looking for software that allows colleagues to post code and data so that others can go through and give comments. This is not necessarially to debug per se, but more so that everyone is using best coding practices (so that projects can be passed off easily) and can learn from other's code.

Eric

[Anders Alexandersson]

Eric, I am not aware of such Stata code review software other than Stata itself. It can be more an issue of workflow and user training than any special software. For example, a good book is "The Workflow of Data Analysis Using Stata". You can use the new user-written Stata program markdoc (GitHub) for "literate programming" if regular Stata comments are not enough but you still want the comments very close to the code.

[Nick Cox]

Perhaps this is redundant, but user-programmers in the Stata world have followed this practice for most of the time since Stata was introduced 30+ years ago.

1. X publishes a program in the Stata Journal [1991-2001: Stata Technical Bulletin] or on a website.

2. Y posts comments and/or code suggestions by personal email or by posting on Statalist. Often, X's program gets revised as a consequence. X often co-opts Y as a co-author; sometimes even promotes them to first author.

Now I guess this looks remarkably primitive to many, but I'd be interested in examples of really good user-written Stata programs written by 4 persons or more. I can think of several with 3 and many more with 2 and many more yet with 1.

I am open to the comment that it's precisely because people aren't using modern software to collaborate that this rapid fall-off occurs;
I am not that dogmatic.

But I've never felt the need for anything like this.

[haghish]

if you need to collaborate on a program, and know what everyone is doing and commenting about the changes they made, and you want it to be free, GitHub is the place to be. Alternatively you can install Git locally, or any alternative. But the Stata community on GitHub is not small and is rapidly growing.