On 7/04/2011 20:37, Ioan Eugen Stan wrote:
2011/4/7 Eric Charles<e...@apache.org>:
Hi Ioan,

I hope your school project is going well.
I have commented your application on google melange (nothing special, simply
a few tips I hope can ameliorate the application).

Yeah, it's going ok.

Glad to hear school project is going fine.

I was a bit nervous and stressed but everything
is going well.

I know that feeling :)


Just want to say also that Robert and I will co-mentor MAILBOX-44.

Tks,
- Eric

I did not receive any comments. Please post them on the mailing list
or on jira. I also noticed my application status is Pending.
I don't know what that means. I also added Robert as a mentor.


Here after the comments I have put on google-melange this morning:

Proposals that you can inject:

- Robert and I will comentor this.

- Illustrate with mailbox modules [1] (you can have picture in application) thatshow you will add the mailbox-hdfs module to the mailbox ones.

- In the design phase, there will be some questions regarding the format and compression (lzo, snappy,...) to be used for hdfs file to store the mails. this will take into account the RFCs mentioned by Robert on https://issues.apache.org/jira/browse/MAILBOX-44

- The design phase will also address the question about the existing lucene implementation we have at James to store the metadatas. Will we use/adapt it,...does it fit the requirements? The current lucene implementation is not distributed.

- The goal is not to have a full distributed james, because we don't have the distributed uid generation (you say it, but it must be highlighted)

- snippet from Robert's mail from which you can inspire from to have a more details on the access layer to hdfs

Though most mailboxes integrate them, there are two distinct aspects
to mail storage. the email content should really be conceptually
immutable, with mailbox and flag meta-data layered on top. I suspect
that the simplest workable solution for MAILBOX-44 (or MAILBOX-45)
would just separate RESTful content storage in Hadoop (or other NoSQL
store) from meta-data retained in an existing data store. This should
be relatively quick and easy to code.

Tks, Eric

[1] http://james.apache.org/mailbox/images/uml/org.apache.james.mailbox-package.png


---------------------------------------------------------------------
To unsubscribe, e-mail: server-dev-unsubscr...@james.apache.org
For additional commands, e-mail: server-dev-h...@james.apache.org

Reply via email to