Aspectify quickstart

[jayvdb]

@satwikkansal had the idea that we might combined aspects and coala-quickstart to create a project that achieves a definite goal without needing to first rewrite everything to be aspectified.

I suspect it would be a solid project to convert https://github.com/coala/coala-quickstart/blob/master/coala_quickstart/Constants.py to using aspects, as it would require more work on aspects, and implementation in relevant bears and quickstart.

[satwikkansal]

I feel like, if well thought out this can be mixed with https://github.com/coala/projects/pull/464. Coming up with the right approach and design might be tricky since the student would be required to familiarise him/herself with coala bears, aspects and quickstart.

[jayvdb]

It would be a steep learning curve, and quite a bit of work, but this concept provides a really clear objective for the project, and sometimes that is the key ingredient to a successful project.

[satwikkansal]

Yeah, a project that connects different pieces of coala/

[adhikasp]

Great idea! @satwikkansal what is your integration goal in coala-quickstart? Something like the default generated .coafile is using aspect setting?

[adhikasp]

Anyway, the aspect part shouldn't be too complicated, because they doesn't need to know the underlying system. Only how to migrate bear and how to write new aspect. So I think a combined project to quickstart isn't too complicated.

[jayvdb]

@adhikasp what do you think about aspectifying https://github.com/coala/coala-quickstart/blob/master/coala_quickstart/Constants.py

[satwikkansal]

Great idea! @satwikkansal what is your integration goal in coala-quickstart? Something like the default generated .coafile is using aspect setting?

Yes, something on the lines of:

• quickstart generates an aspect style quickstart in addition to the conventional one it already generates. We might deprecate the conventional one in future.
• The aspect based .coafile generated is equally (if not more) effective than the conventional type .coafile.
• And few ways to automatically discover the aspects from the meta-data that we collect in the quickstart.

[adhikasp]

@adhikasp what do you think about aspectifying https://github.com/coala/coala-quickstart/blob/master/coala_quickstart/Constants.py

I think its okay. But some of the bears is more tricky to migrate, like https://github.com/coala/coala-bears/blob/master/bears/c_languages/CPPCheckBear.py, because it doesn't directly implement run method. They will need to mess a bit deeper to aspect code/create new helper function for integrating aspect to those bear. Or maybe expanding @linter annotation would works.

[jayvdb]

@adhikasp , ok, so those bears should be left to be a stretch goal for this project.

IMO any bear which cant be aspectified easily should be dropped off the important bear list anyway, as aspect-chosen bears (and settings) will lead to better results.

Anyone who thinks those bears are important can add aspects support ;-)

[jayvdb]

Ah, I see the capabilties system comes from https://github.com/coala/cEPs/blob/master/cEP-0014.md . So yea, this project would involve adapting cEP 14 to work with cEP 5

If I understand correctly, capabilties are disabled by default at the moment in coala-quickstart , as --no-filter-by-capabilities is True.

[satwikkansal]

If I understand correctly, capabilties are disabled by default at the moment in coala-quickstart , as --no-filter-by-capabilities is True.

Yep, because for capabilities we have to ask the user to select that capabilities, and I feel that quickstart already asks too many questions. However, quickstart does drop bears with overlapping capabilities (something that can be re-used with aspects)