Re: [cpp-netlib-devel] URI and shipping 0.4

[Dean M. B. <mik...@gm...>]

Hi Glyn and John.

On Thu, Oct 1, 2009 at 9:24 AM, John P. Feltz <jf...@ov...> wrote:
> I'm glad you posted the link, this basically is the first step of what
> needs to be decided upon for completion of this feature. I think my
> implementation is close in addressing the proposed requirements.

I agree.

> At the moment it's lacking relative resolution and grammar from 1.1 (it's based
> off the 1.0 rfc, I think the differences are minor at the moment). I'm
> desiring an extensible uri parsing solution. While it is a bit of a
> departure from the the original protocol objectives of the library, It
> doesn't veer off too far as to become needless and wasteful.
> Nevertheless, I'm trying not to argue use of my implementation over
> another. The question is ultimately, if the requirements are common. If
> not, than  there are clear reasons to back a particular implementation.
>

I think we can merge the parsers you've built that are closer to the
RFC within the higher level interface that I intend to provide with my
URL implementation.

Extensibility is a matter of design, and the approach I've taken is to
have specific points of extension where different -- and new -- URI
implementations can be provided. I don't believe there is a
one-size-fits-all design for the implementation, and although the
dynamic nature of runtime-generated URI's may require a dynamic
solution (with *ghasp* runtime virtual inheritance). This is only
though if you want to have a URI implementation that can stand on its
own outside of cpp-netlib; where my focus has been for type safety on
a per-protocol basis -- having 'http::uri', 'smtp::uri', or 'ftp::uri'
classes and have clear, type-safe, conversion functions from a
'uri::uri' to the more specific type.

>
> Glyn Matthews wrote:
>> Hi netlibers,
>>
>>
>> I'd like to ask everyone who's working on something for C++ Networking
>> Library to give me a status update.  I notice that some work is being done
>> ondifferent branches and I'm a little concerned that there's duplicate
>> effort going on, specifically for URI processing (I think there are at least
>> 3 or 4 different implementations of URI parsers in various stages of
>> completion).
>>
>> A couple of months ago Dean suggested we start thinking about getting
>> version 0.4 out of the door and I'd like to coordinate our efforts on this,
>> and to release a reasonable quality, generic URI in the near future, so then
>> we can concentrate on improvements and protocol implementations after that.
>>

Thanks Glyn for doing this.

I can only post the user-level API that I expect to provide, which is
shown in the direction that I took when I wrote the unit tests on my
branch.

I think we can definitely merge John's parsers and more RFC-compliant
implementation into my approach.

>> References:
>> https://sourceforge.net/apps/trac/cpp-netlib/wiki/URIAPIRequirements
>>

Should I put the high-level API description in this page too?

Thanks again guys and I hope this helps.

-- 
Dean Michael Berris
blog.cplusplus-soup.com | twitter.com/mikhailberis
linkedin.com/in/mikhailberis | facebook.com/dean.berris | deanberris.com