[MOBY-dev] getting RDF resources in the compressed format

[Paul Gordon]

The question is whether URL.openStream() sends such a header.  Probably 
not, in which case CentralXXXImpl will need to use Apache httpclient or 
something similar instead...

Edward Kawas wrote:
> Okay,
>
> I set up mod_deflate on my local linux installation and tried it out.
>
> Some results:
> Service instance RDF:
> 	Original size: 7052579bytes
> 	Compressed size: 305092bytes
> 	Savings: ~96%
> Object RDF:
> 	Original size: 2864668 bytes
> 	Compressed size: 219478 bytes
> 	Savings: ~92%
>
> I know that it is obvious that we will get big savings by compressing text
> files ... The question was how to implement this ...
>
> Here is what I did:
> 	1. enabled mod_deflate for apache
> 	This differs from installation to installation, but for me, running
> ubuntu, all I had to do was create 2 sym links in /etc/apache2/mods-enabled/
> for the files deflate.conf and deflate.load in /etc/apache2/mods-available.
> 	2. edit the conf file
> 	Again this differs, but what I did was edit deflate.conf and put the
> following inside:
>
> <IfModule mod_deflate.c>
> 	  # mime types to compress
>         AddOutputFilterByType DEFLATE text/html text/plain text/xml
> 	  # support for older browsers
>         BrowserMatch ^Mozilla/4 gzip-only-text/html
>         BrowserMatch ^Mozilla/4\.0[678] no-gzip
>         BrowserMatch \bMSIE !no-gzip !gzip-only-text/html
>         BrowserMatch \bMSI[E] !no-gzip !gzip-only-text/html
> </IfModule>
>
>
> 	3. I restarted apache
>
> More doc available at http://httpd.apache.org/docs/2.0/mod/mod_deflate.html
>
> Now when a request comes in with the header 'Accept-Encoding: gzip' apache
> compresses the RDF.
>
> Eddie
>
> -----Original Message-----
> From: moby-dev-bounces at lists.open-bio.org
> [mailto:moby-dev-bounces at lists.open-bio.org] On Behalf Of Martin Senger
> Sent: February-26-08 4:35 PM
> To: Core developer announcements
> Subject: Re: [MOBY-dev] getting RDF resources in the compressed format
>
> Many thanks, Paul.
>
> I believe the solution we agreed upon was that if the client sends an
>   
>> "Accept-encoding" HTTP header with x-gzip
>>     
>
>
> Yes, that was what I vaguely remembered but was not sure about details.
> [BTW, is there a way to search the whole biomoby email archive?]
>
> How does client know what was sent from the server? Is is in the HTTP
> Content-type header? If so, what mime-type should be there?
>
> Eddie, it would be nice to have this feature implemented on the registry
> site (so other registry providers can copy it, as well). Would you consider
> it please?
>
> Cheers,
> Martin
>
>
>