Consider rebuilding the UI based on CLI arguments/options.

[homeworkprod]

While the current interactive UI seems nice at first, it becomes tiresome when used frequently.

Giving a single wrong value or trying another image requires walking through the whole process again.

Using a command-line interface instead would allow users to re-use and slightly adjust the command they just issued on the shell, and to create very basic shell scripts as some form of configuration file/wrapper.

• Values with useful defaults (e.g. GitHub's hostname) can be assumed and only overridden if necessary, saving hitting <Enter> on every run.
• Values specific to the user's current goal can be re-used and don't have to be typed again on every run.
• An image (or at least one, this could be extended in the future) has to be given no matter what, so why not make that either a positional argument or a required (repeatable) option.

etc., you get the idea.

Then again, CLI arguments could be implemented, but gitfiti could still provide an interactive frontend for newcomers, leaving the former to more experienced, power-hungry users.

[gelstudios]

A lot of those options grew organically, and its a little bit of a mess at this point. I agree that a cli with flag options would be cleaner, but in the spirit of inclusive-ness I'd like to keep the interactive option.

Some defaults like username for calendar can be read from git global settings, etc.

Another issue / usability concern is running gitfiti multiple times against the same repository blows away the generated script on each run 👎.

[homeworkprod]

Adding an option to specify a custom output filename or write to STDOUT would be the obvious fix.

If none is given, the ouput filename could be extended with a (preferrably human-readable) timestamp.