[erlang-questions] Newbie training project proposal

Brian McCallister brianm@REDACTED
Sat Sep 15 01:54:17 CEST 2007


On Sep 11, 2007, at 12:29 AM, Joe Armstrong wrote:

> I'll start with a little rant about the architecture of STOMP :-)
>

Ooo, boy, this will be fun...

> <flame on>
>
> Do protocol designers turn off their brains when they design  
> protocols?

Nope, I think everyone has good intentions at least.

> I see a lot of the following:
>
>    - text protocols with are "easy" to understand and parse

depends on the client, if it is telnet then text is in fact much  
easier. It depends on the goal.

>      (false - most text protocols are inadequately specified)
>

What is inadequately specified?

I am serious, I know the spec plays loosey-goosey with some areas. We  
rely on implementors to let us know where the gaps are. It got a  
"1.0" only because a lot of folks actually use it and no one  
implementing had complained in a while :-)

>   - XML protocols
>     (crazy - did anybody think how long it would take to parse this  
> crap)

Agreed.

> The two most obvious methods for building protocols are:
>
>    - define a packed bit-byte-word structure appropriate to you needs

Not appropriate when a  design goal is that it can be done by hand  
over telnet. A key consumer of stomp is a sysadmin at 3am trying to  
figure out why somethign went wrong. IN that case text format  
protocols are massively valuable. They are also massively less  
efficient, which is why I would generally suggest stomp as a  
transport option on a system and an optimized packed binary transport  
as another.

I think there are glaring crap points in stomp, but the decision to  
be text-oriented is *not* of them :-)

-Brian (who wrote the initial stomp spec because he was tired of not  
being able to access binary protocols easily).




More information about the erlang-questions mailing list