[rrd-developers] Patch: new --compression-level option

Vincent Pit rrdtool at profvince.com
Sat Jul 7 21:18:56 CEST 2007


Hi,

Thanks for the feedback.
> Wouldn't it be better to just pass a pointer to a
> typedef gfx_s {
> 	long width;
> 	long height;
> 	long bbp;
> 	int compression;
> } gfx_output_t;
> instead?
>   
That would surely be better, but it would lead to much more rewrite. I
tried to be as little invasive as possible.
> cheers,
>
> PS: If you're concerned about bandwidth, then i'd use a compressed
> stream to push the images to it's destination. I'm sure there are
> possibilities to use liblzma/libbz2 or the like that are listening on a
> socket. If not, this is a) trivially implemented b) also helps non-PNG.
> Just a thought.
>   
I was essentially thinking about http. Some client/servers support
on-the-fly deflate, but that sounds actually way worse in terms of load
than doing it once in rrdtool.



More information about the rrd-developers mailing list