From: Henrik Nordstrom <hno@ma...> - 2001-05-21 21:19:09
Jeff Dike wrote:
> It would have been more useful to just port that stuff to the new
Agreed, but porting to the new infrastructure is more than the 5 minute hack
needed to get the old running, and I needed to get it running now as I depend
on um_eth_serv while testing my netfilter hacks.
Note: I do not normally need any connectivity with the host as the testing is
done on a virtual network of UML stations. The um_eth_tap tool I use to allow
the host to join the network when needed (mostly for fun actually, as it is
much easier to transfer data using hostfs, which btw all my UML's boot from
using a single shared root fs...).
When I get back to serisous kernel hacking mode (I am currently in maintain
mode only, fully occupied on application level stuff) I will most likely make
TUN/TAP or um_eth_serv transports if it isn't done by then, but from what it
looks I won't be in serious kernel hacking mode for the next month or so, and
until then it will mostly be quick hacks.
> Also, 'cvs diff' doesn't produce patches. Get the two trees on disk and use
> plain old diff for that.
Sorry, should have filtered the cvs diff thru my CVS diff normalizer script to
correct the paths. Normally use scripts to generate (cleaned up) patches from a
heavily branched CVS and is why I sometimes forget when playing manually with
"foreign" CVS repositories..