[mythtv] add support for SQLite?

Isaac Richards ijr at case.edu
Sun Mar 4 18:25:10 UTC 2007


On Sunday 04 March 2007 12:57:24 pm James Chapman wrote:
> Chris Pinkham wrote:
> > * On Sat Mar 03, 2007 at 11:07:51PM +0000, James Chapman wrote:
> >>> Converting is not worth the effort just to safe space/memory.
> >>
> >> I'm not trying to reduce the size of the database either; it's the
> >> filesystem size (i.e. executable/library code size) and run-time memory
> >> footprint that matters to me. Specifically, I'm trying to pave the way
> >> for enlightened commercial STB suppliers to build and sell commercial
> >> STBs with MythTV inside. Right now, the hardware costs are too high for
> >> commercial MythTV-based STBs to be viable. Not enough people would pay
> >> double(?) the price of alternative products just for access to the cool
> >> features that MythTV offers. However, with some tweaks, I'm convinced
> >> that MythTV could be made to fit the hardware price point for deployment
> >> in next generation, commercial STBs. One tweak is to find a database
> >> with a smaller footprint.
> >
> > So doubling flash to 32MB to hold a compressed embedded MySQL along with
> > the extra 32-64MB of RAM is going to double the price of this PVR?
>
> Er, no. :) I was talking about hardware/software costs in general when I
> mentioned the 2x price comparison, not just flash/ram size.
>
> An option to use SQLite instead of MySQL at compile time would make Myth
> more appealing to STB makers. It's not the only thing they're concerned
> about, but it's one of them. Of course, most PC-based users would stick
> with MySQL.
>
> > I'd
> > venture to say that the MPEG-2 license is going to cost more than those
> > two chips combined.
>
> Comes with the hardware system-on-chips that they use. The cost of that
> device, with built-in hardware encoders/decoders/CPU/IR etc, is way less
> than a typical x86/AMD CPU.
>
> > Don't forget about the Qt license as well.  Memory
> > is probably the least thing holding a MythTV STB back, there are much
> > more costlier things that would make up the final price.
>
> That's why I used the phrase "enlightened STB makers" - if the STB
> software that uses Qt is open source, there's no license. :)
>
> Anyways, the reason for my original post was to find out if anyone was
> already working on SQLite so as to avoid doing duplicate work. It seems
> no. So I'll go away and develop a patch.

We don't really want to have to support multiple database types.  It's quite a 
bit of extra work.  I _would_ like to switch to embedded mysql - that should 
get us most of the largest benefits of moving to sqlite without the added 
pain of reworking large amounts of code.

If you're interested on working on that, instead of sqlite, I can write up 
some quick docs for you on how I think things should work.  Or, just try to 
find me on IRC (Chutt in #mythtv on freenode) and we can talk about it there.  
I think you were on earlier, but I wasn't around then.

Isaac


More information about the mythtv-dev mailing list