> Al and Peter have asked me to allow the individual usb-serial drivers to > become separate loadable modules (like the other usb drivers with the added > need for the usb-serial core module to also be loaded.) > > To do this would only require a small change to the usb-serial code > > Here's the pros and cons of this as I see it: > Pros: > - enable vendors to provide usb to serial drivers outside of the kernel > source tree easier (Digi would like this, as it enables them to update > their drivers to their customers easier, and their drivers are GPLed.) > - reduce the memory size of the kernel footprint if you are only going to > use a device that requires a firmware download every once in a while, > but are going to use your Visor all the time. I'd still consider memory cheap these days for any system capable of supporting a usb controller. Making them seperate modules has my vote. > Cons: > - enables vendors to provide usb to serial drivers outside of the kernel > source tree easier (I know, this is also in the pro list, but this > change would allow vendors to ship binary drivers that play nice with > with the usb-serial major:minor numbers as they really can't do this > today very easily.) I think the best answer is to allow for this; vendors that won't release source won't release a driver at all if they can't release a binary one. If a vendor wants to ship binary only drivers for a serial adaptor, let them, they're likely to have to invest -more- work on their driver because they'll have to do the porting/testing/maintenance/bugfixes themselves for each new kernel release. Let vendors learn by watching which hardware Linux users are more likely recommend and buy (the ones with good free drivers). > - requires users to load yet another module to support their device (to > use the visor would require both usb-serial.o and visor.o to be loaded, > although this can be taken care of by the modutils package easily.) That's what modprobe is for. mobprobe visor should load usb-serial first. Greg PS I hope your move goes well -- Gregory P. Smith gnupg/pgp: http://suitenine.com/greg/keys/ C379 1F92 3703 52C9 87C4 BE58 6CDA DB87 105D 9163 --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
