Re: [ros-users] getting a topic's md5sum without subscribing

On Thu, Apr 1, 2010 at 3:07 AM, Ken Tossell <ktossell@umd.edu> wrote:
> On Mon, 2010-03-29 at 14:46 -0700, Josh Faust wrote:
>> Do you not get warnings when doing that?  I'm pretty sure we've been
>
> I do get that warning.
>
>> What output do you get when they mismatch?
>
> I was mistaken about this -- if the messages have roughly compatible
> fields (say, "string a; string b" -> "string a; string c"), then the
> message goes through with no warning. If they're incompatible (say,
> "string a" -> "string a; string b"), then the connection is dropped
> silently.

There should be a warning if the md5sums do not match (e.g. in
rxconsole). If there isn't, there's a bug worth ticketing.

> Is there a reason why the master tracks the message type but not the MD5
> hash or the message definition? I imagine having that information in the
> master could be useful (prevent two publishers with different message
> versions both publishing to a single topic).

The master doesn't track the md5 sum because there's nothing it can do
about it. It can't arbitrate who's right, and the publishers and
subscribers have to validate each other regardless. Basically,
end-to-end principle.

- Ken


> Ken
>
> _______________________________________________
> ros-users mailing list
> ros-users@code.ros.org
> https://code.ros.org/mailman/listinfo/ros-users
>