Hi all,
Back from my vacation trip and I see quite exciting things have 
happened in these 10 days!

2006-08-22: EV dixit:
> Unfortunatelly, I am neither really content with my current
> solution of accessing lkarmafs-uploaded virtual files from
> within lkarmafs itself. [...] So I am changing my mind and
> going back to the temporary file solution [...] a head-tail
> copy of the whole file; just enough data for libtag to work
> properly.  Since this does not allow computing the rid, [...]
> my current idea is to compute the rid incrementally in [...]
> kfs_write()

2006-08-23: Keith Bennett dixit:
> [...] I had a quick look at this last night and it seems that
> you are out of luck. The algorithm makes use of the md5 sum
> from half-way through the file. In lkarmafs there is no way of
> knowing where that is until after you've finished writing the
> file.

2006-08-23: Frank Zschockelt dixit:
> I'm not sure whether my idea is usable, but the Pearl Function
> "Read File Chunk" has an offset. So my idea would be:
> 
> 1. Upload the file via ethernet
> 2. Read the 192kb for the rid-calculation with
>    lk_karma_read_file_chunk
> 3. Calculate the rid
> 4. Updating the properties

2006-08-24: Keith Bennett dixit:
> I wasn't aware of that. It is an excellent idea. We will
> already be saving the head and tail in order to extract tag
> information, so we just need to read the 64k from the middle of    
> the file. That's bound to be miles quicker than any other
> method.

Indead, this is a very good idea -- likly a perfect solution!

2006-08-30: Keith Bennett dixit:
> This patch implements the lk_rio_update_props_from_tags() which
> can be used by lkarmafs to update tag details for both USB and
> Lan attached devices. This version no longer passes a filename
> argument as the library does everything itself.
>
> The patch should apply cleanly to hg112 (ie. after all my 
> previous patches are applied). There is also a patch for
> lkarmafs which is against the vanilla 0.1.8.
>
> If a USB device is in use, then the file on the karma is used 
> directly to obtain the tag information. If a Lan device is
> used, then a temporary file is created under /dev/shm (or /tmp
> if that fails) containing just the header+64k, mid 64k and
> tail+64k. These are obtained from the karma using the
> lk_karma_read_file_chunk() as per Franky's suggestion.

Fine work, thanks.  After a quick look, the patches seem good.
 
> Note to Franky: you should probably hold off on applying this 
> patch until EV gives the go-ahead. If libkarma has changed by
> then, I will post a new version.

I'll make some tests this evening and I'll let you know.

Cheers,
EV.


-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
linux-karma-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/linux-karma-devel

Reply via email to