Hi Fabien,
I completely agree that it's not a high priority. Just noting it as I'm trying to convert to using Koneki LDT. :)
As for the inferring of type information from code - assuming you mean from C/C++ code, is it really worth it? I mean the API developer will have to write the help/description text anyway, which is a big chunk of the comments to write. Although having said that, I agree it's usually the param/return types that are wrong in the comments/documentation, rather than the description, because the developers copy/paste the comments and change the description but forget to verify the types... hmmm, maybe it is worth
it.
-hadriel
From: Fabien Fleutot <fleutot@xxxxxxxxx>
To: Hadriel Kaplan <hadrielk@xxxxxxxxx>; General development dicussions for the Koneki project <koneki-dev@xxxxxxxxxxx>
Sent: Thursday, March 1, 2012 4:47 AM
Subject: Re: [koneki-dev] The Record-block/Type-block
On Thu, Mar 1, 2012 at 8:59 AM, Hadriel Kaplan
<hadrielk@xxxxxxxxx> wrote:
But due to poor design [...] some APIs have many different enum lists/types they put in the same sub-table/global.
Indeed, there are scary stuff on the wild :) However I'd argue that:
- given how ambitious the project is, at the very least supporting such poor designs shouldn't be a priority;
- if writing correct code enables better IDE support, then a couple of people will start writing better code, and the world will be a slightly better place;
- it would be very great if, in the future, a lot of typing information was inferred from the code rather than mandatorily written down in the comments. The messier the type system, the harder inference becomes. I'd rather leave the door open for inference than help people write unidiomatic code!
