I’ve gone through all the pages for the Git course. Now, I’m doing another read through with a focus on creating distinct, regularly reviewable tasks. I’m utilizing a dynamic review process that adapts based on the degree to which I get something right. This enables me to conduct reviews quite efficiently, but also results in a nonlinear review path.

For instance if I have a hard time with doing something which comes up after multiple prior steps building up to that point, I don’t want to have to constantly redo the steps I understand. Obviously, for the exercises with a command line focus, I can write command line scripts pretty easily to deal with dynamically generated review content.

For the things that talk about how to do things in Eclipse itself, though, it would be extremely useful to be able to have programmatic control there too. Launching with a specific workspace seems easy, I just use the -data command line option, but it seems like after that things become very opaque, and Eclipse turns into what is sometimes called a CUI (Captive User Interface).

In theory I could solve this by setting up a unique, manually generated preset for each of my review cases, but that seems super clumsy, very time consuming, and would put me in a situation of needing to update them all manually as Eclipse changes. It would also become a progressively worse situation as the scope of what I want to review grows, thus disincentivizing me from striving to be able to do more with Eclipse.

Is there an alternative to that super frustrating manual treadmill? Some way of programmatically triggering wizard page settings, next button clicks, finish clicks, hotkey presses, etc? An Eclipse "remote control"? Obviously a programmatic approach would also need to be revised as Eclipse changes, but it would present the opportunity of being done in a modular fashion where changes could be limited to common base modules rather than taking a presets approach where ALL the presets would have to be revisited one by one if Eclipse changes something of broad effect. It seems like this will be an even bigger deal when I get to the same point with the RCP course.

As a peripheral benefit, such capability would also make it possible to submit very precise issues with regards to reproducibility, as it would allow a programmatic submission potentially all the way down to the workspace directory creation and launch point.