bash or Python? The Square Pegs and a Round Hole Situation

The question “should I do it in bash or in Python?” is both frustrating and common. Why one even needs to choose between two alternatives which are both inadequate for the task at hand? Why try to pick one of the square pegs for the round hole? I believe that you should not be in this annoying situation when just trying to write a script and get back to your endless stream of other todos.

Best illustration that I managed to find in 2 minutes

Both are Inadequate for Ops

bash does not meet any modern expectations for syntax, error handling nor has ability to work with structured data (beyond arrays and associative arrays which can not be nested). Let it go. You are not usually coding in assembly, FORTRAN, C, or C++, do you? They just don’t match the typical Ops tasks. Don’t make your life harder than it should be. Let it go. (Let’s not make it a blanket statement. Use your own judgement when to make an exception).

Python along with many other languages are general purpose programming languages which were not intended to solve specifically Ops problems. The consequence is longer and less readable scripts when dealing with files or running external programs, which are both pretty common for Ops. For example, try to check every status code of every program you run, see how your code looks like. Sure you can import 3rd party library for that. Is that as convenient as having automatic checking by default + list of known programs which don’t return zero + convenient syntax for specifying/overriding expected exit code? I guess not.

Your disorientation and frustration is completely legitimate.


Multitude of attempts to provide viable alternatives by different people are in progress. As others authors, I would like to help my Ops colleagues to avoid frustration and be productive. It just feels good.

Informative section is over. Shameless plug about an alternative that I am developing and how it is special follows.

Next Generation Shell as an Alternative

How Next Generation Shell differs from the alternatives? Reasonable question that I would ask too before investing any more time if I was the reader.


Current shells as well as proposed alternatives treat UI as nothing has happened since the 70-s: mostly typing commands and getting some text back.

How about real interaction with the objects on the screen? Oops. In a typical shell there are no objects on the screen, it’s just a dumpster of combined text (if you are lucky; could be binary) from stdout and from stderr from one or more processes (unless steps were taken). WTF? It doesn’t help you to win. It helps you to lose time. NGS does what is intended to make you productive. I have organized my thoughts about how the UI should look and behave on the wiki page.

Programming language

It looks like alternative solutions have the “let’s make the shell better” approach and therefore are heavily based on the shell syntax and paradigms.

There is also “let’s make a library for existing language” approach which doesn’t hit the target either. Can’t have a syntax for common ops tasks for example.

And finally there is “let’s make a library and a syntax on top of existing language”. Not sure about this one. Sounds good in theory. Looked some time ago at something like this and the overall impression was … awkward (for the lack of better word).

Approach in NGS: let’s make a good programming language for Ops, which fits the use cases and has syntax and facilities for the most common tasks such as running external programs.

The language follows the principle that the most common tasks should have their own syntax or a library function (depending on usage frequency). Examples:

  1. ``external program`` – runs external program and parses the output (JSON is auto detected, easily extensible for anything else).
  2. status(), log(), debug(), retry() – standard library functions. How many times an Ops person should write his/her own retry()? It’s insane.
  3. Argv() facility for constructing command line parameters (for calling external program).
  4. p=$(my_prog my_args &); ....; p.wait().

Small number of “big” core concepts in the language (types with inheritance, multiple dispatch and exceptions).


Most of the standard library is in NGS.

The UI (only recently started working on it) is in NGS. It doesn’t make sense that when a user of the shell wants to fix a bug in the UI and suddenly he/she needs to learn Go, Rust, C or whatever other language.

We use NGS at work.

Most of the demo scripts come from either current or previous work.

How to proceed?

  1. Install NGS.
  2. Consult documentation and look at sample scripts
  3. Write scripts for non-production-critical tasks.
  4. I am here to help. Do not hesitate to contact me with questions, suggestions, or feedback. If there is anything Ops-y you are trying to do seems to be easier in bash or Python – open an issue, because that’s a bug from my perspective. Something is inconvenient? Yep, also a bug.

If you are like me, you will find at least some satisfaction in using the most appropriate tool before continue to the myriad of other tasks that are in your todo queue.

Or Just Learn More

  1. Is NGS for you? Take a look at intended use cases.
  2. Take a look at how NGS compares to other programming languages.
  3. Browse sample scripts to get some impression about the language.