So I have done some more work on my CSV parser. I haven't gone to any effort of benchmarking the two implementations, but I think what is important is the interface and that it works with an input range.

My question is what should I have to be ready for a review? Should I make a pretty Phobos like documentation available? I don't think I have it very well documented but think the Review process is a great time to get feedback on this so I know where I'm lacking.

I look forward to waiting in the review queue.

-- Jesse Phillips

On Tuesday 02 August 2011 21:33:05 Jesse Phillips wrote:
> So I have done some more work on my CSV parser. I haven't gone to any effort of benchmarking the two implementations, but I think what is important is the interface and that it works with an input range.
> 
> My question is what should I have to be ready for a review? Should I make a pretty Phobos like documentation available? I don't think I have it very well documented but think the Review process is a great time to get feedback on this so I know where I'm lacking.
> 
> I look forward to waiting in the review queue.

Ideally, the module would be at the point that if its design and implementation were perfect, it could be immediately merged immediately into Phobos. So, it should be as good as you think that you can reasonably get it prior to the review, and it should be fully documented. Obviously, the review will help iron out any problems that it may have, but I believe that the idea, at least, is that it be as ready as you can make it prior to the review.

Now, if you're still trying to figure out what to do with your design, then it would probably be best to put it up for an informal review process in the main newsgroup (both std.log and std.path have done this). Whatever feedback you get from that can help you iron out whatever you need to iron out before deciding on what API or implementation to go with.

But I would say that if you don't have what at least _could_ be immediately merged into Phobos, it shouldn't be put up for formal review and voting. Naturally, some reworking of the design and implementation is bound to occur in the formal review process, but I think that the formal review process is more for refining the design and sorting out any issues that you missed rather than being for deciding on a design. Informal reviews can be used to sort that out first.

As for what's next in line for review, it's probably either the curl wrapper or std.log, depending on their exact status and the status of their implementors.

- Jonathan M Davis