[Rprotobuf-yada] bytes

[eckuipers-web at yahoo.com]

Hey Dirk & Romain, 
I just read both your replies. Thanks for getting back so quick. It's okay that 
i missed the release, I will just use the SVN if/once you decide to switch to 
raw.

Getting to the raw bytes is particular important for RPC implementations, since 
they almost all rely on this feature. This is how i ran into it: i was trying to 
build a R client for my protobuf over zmq RPC implementation (soon to be 
released open source).

To be specific: in RPC it is up to the channel implementation to get both the 
request protobuf message and the function name across to the server for the 
request. And for the response the server has to send back the response protobuf 
message and possible errors. Naturally one wants to use protobuf for this. But 
since the request message and response message are unknown classes from the 
RPC's point of view, the only way to deal with them is as serialized raw bytes. 
So most RPC implementations have their own proto file that looks something like 
this:



message Request {
  required string method = 1;
  required bytes request = 2;
}
 
message Response {
  optional bytes response = 1;
  optional string error = 2;
}
Best,
Koert


----- Original Message ----
From: Dirk Eddelbuettel <edd at debian.org>
To: eckuipers-web at yahoo.com
Cc: rprotobuf-yada at lists.r-forge.r-project.org
Sent: Fri, September 10, 2010 6:54:46 PM
Subject: Re: [Rprotobuf-yada] bytes


Hi Koert,

Thanks for posting on-list! 

On 10 September 2010 at 15:19, eckuipers-web at yahoo.com wrote:
| Dirk, Romain,
| Why did you decide to represent the bytes field type by an R character vector? 

| Why not raw?
| I have a message with a bytes field, and within R i cannot get to the bytes 
| since the character representation cuts it off at the first null character it 
| runs into, it seems. So charToRaw doesnt help me here either. 
| 
| Any suggestions?

Fair question. 

We just happened to have uploaded 0.2.0 which should be on CRAN tomorrow --
did you check against the ancient version on CRAN or did you check against
SVN?

Dirk

-- 
Dirk Eddelbuettel | edd at debian.org | http://dirk.eddelbuettel.com