MXP Message Draft 0.6

12 views
Skip to first unread message

Tommi Laukkanen

unread,
Dec 10, 2009, 12:49:44 PM12/10/09
to Metaverse eXchange Protocol
Hi

I have translated the mxp message definitions to .proto syntax of
Google protocol buffers. The agreed changes have also been
incorporated but it would be good if someone doubled checked the
decisions against the draft:

http://www.bubblecloud.org/wiki/-/wiki/Main/MXP%20Messages%20Draft%200.6

There are no comments yet but just the plain .proto syntax.

-tommi

Maróy Ákos

unread,
Dec 14, 2009, 6:03:45 AM12/14/09
to metaverseexc...@googlegroups.com
Tommi,
Some initial questions / remarks:

- naming convention: MVector3f and others: why the M prefix?

- naming convention: I see you're using Pascal case, like FooBar, but
wouldn't camelCase be nicer:
http://msdn.microsoft.com/en-us/library/x2dbyw72(VS.71).aspx ?
so that for for example for vectors, we'd have x, y, z, and not X, Y and Z?

the lower-camel-case version (e.g. lowerCamelCase) would be a prefered
way in Java, while C++ would prefer lower_camel_case instead. but still,
if the members start with lower case, that's already better IMHO :)

also, the google protopbuf page uses / suggests lower case member
names:http://code.google.com/apis/protocolbuffers/docs/overview.html

for types, the upper case CamelCase should be OK I guess


otherwise, it looks good..


Akos

Tommi Laukkanen

unread,
Dec 14, 2009, 6:44:32 AM12/14/09
to metaverseexc...@googlegroups.com
Hi

Good point we need to follow the google protocol buffers convention.

Regarding the MVector3f etc we need to try to avoid using same names as the math and 3d libraries use in different platforms or the conversion code needs to refer them with namespace inline. In general making identical class names with well known libraries you will be using in tandem with your own code is bad in my opinion. But I guess this is matter of taste as namespaces exist to solve this kind of problem. We could use simple Vector3f as well or even vector3f.

-tommi

Reply all
Reply to author
Forward
0 new messages