于 2013/4/17 13:47, Antony Polukhin 写道:

Hi, and welcome!

Start from writing a proposal according to this link https://svn.boost.org/trac/boost/wiki/SoCSubmissionTemplate . This link may also help https://svn.boost.org/trac/boost/wiki/SoCHints

After it is done, mail it ti this mailing list with title starting from [boost] GSOC. I also recommend to read this mailing list, because there is a lot of interesting discussions and useful information about GSOC.

-- Best regards, Antony Polukhin

_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

Thanks very much, do you mean that I should submit a proposal to the list before the GSOC student application opens?

On 2013/4/18 15:19, Antony Polukhin wrote:

2013/4/18 Hardy :

...

Thanks very much, do you mean that I should submit a proposal to the list before the GSOC student application opens?

Yes, you shall do it before the 21st. There are too many proposals, all of them are interesting and usually require a lot of discussion. Library maintainers and GSOC Mentors will be asking you about your experience, how are you going to implement your proposal, giving advices, arguing with each other... All of this require some time. After that - best proposals (not all!) will be accepted. More discussion - higher chance for proposal acceptance.

-- Best regards, Antony Polukhin

_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

Thanks for your reply, it is crucial information for me. Since I found that the ropes data structure didn't seem to have an implementation in boost(there's a rope_traits in it, but I don't think it is an implementation, so I most wanted to make a proposal on it. Further more, the algorithm on idea page, radix sort and data structure, trie is also an good and confident choice for me. could you please give me some suggestion about it? Thank you very much!

On 2013/4/20 18:21, Bjorn Reese wrote:

On 04/20/2013 05:14 AM, Hardy wrote:

...

Further more, the algorithm on idea page, radix sort and data structure, trie is also an good and confident choice for me. could you please give me some suggestion about it?

I once wrote a radix tree (or trie.) It was a template class, just like the other STL containers, which took three template arguments: key type, value type, and scanning policy. A custom allocator could be added, I just didn't have the need for it.

The scanning policy is used to parse the key, and I wrote two of those. The first was a character scanner, which simply breaks the input string into a sequence of characters, and as such it resembles the normal trie.

The second was a namespace scanner, whose input is supposed to be namespace string (e.g. "Alpha::Bravo::Chalie") and which breaks the input into a sequence of namespace components (e.g. "Alpha", "Bravo", "Charlie".)

It would be easy to supply other scanning policies, such as dot separated strings (e.g. "Alpha.Bravo.Charlie" as used by default in Boost.PropertyTree), paths (e.g. "Alpha/Bravo/Charlie"), and even UTF-8 character scanning.

_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

Thanks for your information on trie, it's useful for me when considering what I should choice for my proposal.