[mythtv] Intro and Mythweb

[Kevin Kuphal]

Isaac Richards wrote:

>On Tuesday 01 November 2005 10:39 am, Kevin Kuphal wrote:
>  
>
>>Isaac Richards wrote:
>>    
>>
>>>On Tuesday 01 November 2005 09:37 am, Kevin Kuphal wrote:
>>>      
>>>
>>>>Chris Petersen wrote:
>>>>        
>>>>
>>>>>>Although this could certainly done in Mythweb itself, another
>>>>>>approach could be to extend the mythtv protocol to provide some
>>>>>>program guide access functions,
>>>>>>            
>>>>>>
>>>>>I'm *all* for this.  personally, I'd much rather see the protocol
>>>>>expanded so that clients don't have to talk directly to the database.
>>>>>Unfortunately, that's not how it currently works.
>>>>>          
>>>>>
>>>>I would whoheartedly agree here.  Having everything in the protocol
>>>>eliminates the possibility of a client doing something wrong to the DB
>>>>because all calls are made through the backend which will properly
>>>>handle the call.  And if nothing else, it eliminates the need for a DB
>>>>client and configuration on the client.
>>>>        
>>>>
>>>frontend -> backend -> dbserver -> backend -> frontend
>>>vs.
>>>frontend -> dbserver -> frontend
>>>
>>>If you were a simple query, which would you prefer to do? =)
>>>      
>>>
>>Oh, I agree.  There is certain amount of gain in going from frontend to
>>db.  The question remains, is that gain worth the increased complexity
>>in having multiple frontend platforms having to rewrite code to perform
>>the same task (perhaps badly) as well as the configuration necessary to
>>support it.  How many times have we seen users hit by trying to
>>understand configuring MySQL for access by remote frontends?  To me it
>>is the same arguments people use when evaluating proxy servers for web
>>access.   And yet, it is quite comman and accepted to use them because
>>the gains outweigh any penalty incurred by the extra step.
>>    
>>
>
>A proxy server cuts out (potentially very slow) external network access and 
>repeated identical queries.  You wouldn't use a proxy server for local 
>content, I don't think.
>  
>
I suppose.  You could show benefit from the caching mechanisms and I 
think in our case, the code reuse between frontend platforms is an 
excellent opportunity for improvement, especially with projects like 
MythWeb, MythRoku, MythMVP and others coming online. 

Making the protocol more inclusive by adding additional commands does 
not disallow direct DB access by the frontend, but provides a standard 
mechanism for other frontends to speak to the backend, especially for 
platforms where MySQL access is unavailable or undesired. 

Bottom line is that it is your project and I certainly resepct your 
decisions with regard to something as low level as the protocol. 

Kevin