DConf hackathon: idea list

[Adam Wilson via Digitalmars-d-announce]

On 5/4/17 16:33, rikki cattermole wrote:
> On 04/05/2017 3:22 PM, Adam Wilson wrote:
>> On 5/4/17 15:32, Seb wrote:
>>> Hi all,
>>>
>>> the DConf hackathon isn’t a hackathon in the traditional sense. It is
>>> intended as a day for _collaboratively_ focusing on long-lasting
>>> problems and pain points in the D ecosystem, planning upcoming features
>>> or DIPs, and creation of a rough roadmap for the next months.
>>> Of course, any D hackers who wish to simply progress their own personal
>>> projects are welcome too!
>>>
>>> Experience has shown that in large groups too much time is wasted on
>>> giving a voice to everyone, whereas for tiny groups chances are that it
>>> takes too long to get the ball rolling.
>>> Hence, a group size of four or five D hackers is recommended.
>>>
>>> Below you can find a list of themes with a short abstract and a couple
>>> of ideas. The abstracts and ideas are intended to get you started and
>>> guide you. Please feel free to _add your own ideas_ and _add your names_
>>> next to them so that people can ping you (IRC, email, and other IM
>>> handles might be handy as well). Of course, you can add your name to
>>> multiple projects.
>>> On Sunday the first half an hour will be used to finalize the group
>>> forming. All existing groups and persons with an idea, but without a
>>> group can pitch their idea shortly (one minute max, no slides) and thus
>>> find other motivated D hackers.
>>>
>>>
>>> https://docs.google.com/document/d/1L5edu6LLj3Afa3tPgqk-aX-fErwr7sPj37Dt5avoc5w/edit#
>>>
>>>
>>>
>>
>> From the Phobos wishlist:
>> I am working an a generic SQL database interface. If anybody is
>> interested in helping out I have a small amount of code that shows the
>> general design direction I've taken so far. We can discuss the design
>> and collaboratively hack out a prototype.
>>
>> The current code is here:
>> https://github.com/LightBender/std.experimental.database.sql
>
> Looking at that I think focusing on describing of tables ext. would be a
> good first step. I worry that it won't be very flexible memory
> management or serialization wise.
>

It's definitely the hardest problem to solve. Obviously we could start 
with a brute force approach (box everything like ADO.NET does) for the 
sake of getting started and then work on improving it later. For now I 
compromised and used Variant, which should be sufficient for most cases 
right now. (See latest commit) I am using Classes and Dynamic Arrays so 
it's not @nogc yet, but this is something that can be improved over 
time. There is no reason that this could not use the new Allocators API 
at a later date.

I'm not looking for inclusion into Phobos during the DConf Hackathon, 
but I would like to block out something that we can start using.

-- 
Adam Wilson
IRC: LightBender
import quiet.dlang.dev;