Remove "View-based architecture development"

[mikesperber]

I've finally read the sources we cite [Rozanski+2011] and [Kruchten 1995] (or rather, a summary paper of Rozanzki).

• [Rozanski+2011] is very explicitly about Architecture Documentation: There's no description of a development process.
• [Kruchten 1995] does describe a development process, but it's called "A scenario-driven approach", i.e. not based on the views - the views only document the architecture.
  Consequently, I still have no clue what to say about this in a training. (Or how to formulate an exam question.)
  Both [Rozanski+2011] and [Kruchten 1995] are good sources, but we cite them again in LG 04-05, so nothing is lost if we delete this item.

[alxlo]

You are referring to LG 03-03, right?

I think you're correct that neither [Rozanski+2011] nor [Kruchten 1995] explicitly describes a view-based development process. I'd argue the underlying idea is worth keeping, just better contextualized.
My take: developing the views is developing the architecture. They're not a post-hoc description of decisions made elsewhere. So how about reformulating it as:

view-based architecture development (developing views as the primary design activity, not merely as documentation), see [Starke 2024], [Rozanski+2011]

Swapping [Kruchten 1995] for [Starke 2024] also makes sense here. Arc42 embodies this thinking most directly, and Kruchten isn't lost since he's already cited in LG 04-05.

[mikesperber]

You are referring to LG 03-03, right?

Yes.

view-based architecture development (developing views as the primary design activity, not merely as documentation), see [Starke 2024], [Rozanski+2011]

Swapping [Kruchten 1995] for [Starke 2024] also makes sense here. Arc42 embodies this thinking most directly, and Kruchten isn't lost since he's already cited in LG 04-05.

I just checked [Starke 2024], which has a chapter on "Design" - view-based architecture development is not there.

I'm totally willing to believe that view development could be a primary design activity, but I have zero basis for figuring out how that would work from the literature. This creates a problem for trainers, especially as this LG item is R1.

[alxlo]

Would a glossary entry be enough to clear up what is meant by this (sorting architectural design artifacts into views and itersating between views) or should we aim higher?

[mikesperber]

Well, a glossary entry would help, but arguably, this needs a more prominent place, alongside ... "Don't call it quality requirement". Maybe the glossary needs a "white-paper section" on topics of special interests to trainers?