On Monday, 10 September 2018 at 00:56:37 UTC, void wrote:
On Sunday, 9 September 2018 at 06:32:39 UTC, Jonathan M Davis wrote:
On Saturday, September 8, 2018 8:36:26 PM MDT void via Digitalmars-d wrote:
On Sunday, 9 September 2018 at 01:30:14 UTC, Neia Neutuladh wrote:
> On Sunday, 9 September 2018 at 00:20:04 UTC, void wrote:
>> [...]
>
> https://code.dlang.org/packages/index.json
>
> https://code.dlang.org/api/packages/[package name]/info

Thanks.

I downloaded 100 packages from code.dlang.org and search for "*doc*" and "*example*"

The results:

13 packages with "*doc*"
41 packages with "*example*"
55 packages with neither
8 packages with both

What would you expect that to tell you? ddoc doesn't require any kind of doc folder (though some projects would have one for custom ddoc files that then affect how the documentation looks), and usually, the best way to handle examples is to use ddoc-ed unittest blocks. So, while in some cases, it would be appropriate to have some sort of additional documentation or examples separate from the source code, in general, there isn't much need for it. So, I don't know what information anyone would expect to get from whether or not a project has any folders with doc or example in their name.

- Jonathan M Davis

Script updated now searches for the string "unittest" on package directory (*.d files only).

Result:
48 packages with "unittest"

That means there is ~= 50% chance an user should read the library source in order to use it. To be fair I should manually check if README.md is decent enough.

I'm sorry, but it's not even close to accurate, because some libraries has documents on additional websites that has examples and no examples directly in the source code using "standard unittests" - This is true for most big libraries / frameworks.

You can't really measure such a thing accurately.

Reply via email to