Recommended Approach for Pliny App with no DB

[naaman]

I just created a small service that has no need for a database; the service uses REST APIs as its backend. To clean up dependencies and operations, I had to remove a few things from the generated pliny app:

• db related gems, pg and sequel
• a handful of config
• the db folder
• pliny tasks that try to run db tasks by default

Does it make sense to provide a --no-db option for pliny-new? If so, what else should needs to be avoided? If not, what's the recommended approach for cleaning up a new pliny app that doesn't need a database?

edit:

Also need to remove

• DatabaseCleaner in spec helper

[pedro]

+1. We shouldn't go crazy on these pliny-new options (eg: no --use-activerecord 😬 ) but no db is definitely a use case we've seen before.

[brandur]

I'm a light +1 as long as we can implement the new pliny-new path in such a way that it's unlikely that we'll see any major regressions.

Generally, it seems like a useful enough feature, but I would guess one that's seen infrequently enough that we could probably get away without a special installation path if no one's too keen on tackling the patch.

[naaman]

Yea, I don't know what the general usage is, but it's the second time I've had to do it (in ~1 year). If I get around to it (we're on a dreamforce tear atm), I might submit a patch and we can mull it over from there.

That said, anything glaring I'm missing from the list above? I think the one thing I forgot to mention was removing the DB transaction stuff from the specs.

[brandur]

@naaman Ah man, sorry for the late reply. My email etiquette's been terrible lately.

That said, anything glaring I'm missing from the list above? I think the one thing I forgot to mention was removing the DB transaction stuff from the specs.

I think you mostly got it. I just took a look through the templates folder and a few others jumped out at me:

• models/ + associated spec folder.
• migrations/.

I think that's it though.