Drasil: Automate SysCntxt

[samm82]

From #1602. The SysCntxt section should be automated.

[samm82]

In all examples, the only external entity is the user, but this could change in the future. Should this line be automated by passing a list of the external entities, or should it be removed altogether (since the information is readily available in the diagram itself)? It could also be generated to always be the user, but I don't think that would be a good long-term solution. This line is currently present in all examples except for SSP. @JacquesCarette @smiths EDIT: I personally think it should be removed, as it doesn't really provide any additional useful information.

[JacquesCarette]

Don't over-automate. We should generalize when we have specific, concrete examples that need it.

[samm82]

So it should always say "the user in this case"?

[JacquesCarette]

I wasn't referring to that - that wording is really something @smiths should make a call on. I was being more generic.

[smiths]

@samm82 this is a great discussion, but I think it is premature to over-automate in this case. I agree with you that the external entity might not always be the user; it might be another software system. I'm not sure how often this will come up, but we definitely don't want to "hard-code" in user in all the examples. Since we know this is a potential point of variability, but we do not yet fully understand the potential parameters of variation, I think we should leave this alone for now. There are definitely higher priorities for automation, and higher priority areas for deepening our understanding.

Since this is labelled as future enhancement, I think we can keep the issue for when we are ready to deal with it. I don't think it will be something we'll get to this summer.

[samm82]

I had started work on this in the sysCntxtAuto branch (and had forgotten about it). Should I try to finish it up or should I not worry about it? I think I might go ahead and try it as best as I can with just me making the design decisions, and then let you review the PR and see if it's worth merging. @JacquesCarette @smiths

[smiths]

@samm82, I don't want to micro-manage you at this point in the summer. If you think you are on to something good, I'm fine with you giving it a try. However, this is far down on my personal list of priorities. I think the starting point would be a domain analysis of the commonalities and variabilities of "system context" for scientific computing software. Once that was done, we could think about how to automate. As interesting at this would be, I think the efforts on automation should at first be focused on artifacts that interest most potential users, and the system context is something that almost nobody currently documents at all. (They should document it (in my opinion), but they tend to leave it implicit.)

[samm82]

Should I just delete the branch then? (It wasn't too much work to do)

[smiths]

I think we can safely delete the branch, but maybe @JacquesCarette should have a look first?

[JacquesCarette]

As far as I can tell, that branch contains just one day's worth of work. This is low priority enough, and likely enough to conflict with other things we want to do in this direction, that I doubt it's worth keeping.