Greg, Thank you very much for starting from my /proc document and for posting the sensors-sysfs document for comments. Do you have an updated version to send out? I see you posted it on 3/27 but got a lot of comments after that. I would like to comment but seems like I should start from the latest. thanks mds Philip Pokorny wrote: > Greg KH wrote: > >> On Sun, Mar 30, 2003 at 02:47:24PM +0200, Martin Schlemmer wrote: >> >>> This should have a working w83781d driver updated for 2.5.66-bk4. >>> Currently sysfs support is working, and are according to the spec >>> (sensors-sysfs) in the 'lm sensors sysfs file structure' thread. >>> Thus I used 'temp_input[1-3]', as there was not final decision >>> on having them temp_input[0-2] as well, for example. > > > Hey... > > Just thinking, about this some more. > > Can you still use sysctl() against sysfs values, or is that also > deprecated? Could a sysctl call return the value and magnitude where > the proc/sysfs interface did a decimal point conversion of the value? > > What are your thoughts on creating *directories* for each sensor and > putting the multiple values associated with that sensor in the > sub-directory. > > Thus, instead of in0, in0_max (or in_max0), etc. we would have > in0/current, in0/max, in0/min, in0/offset. > > My thoughts are > * This keeps the directory size a little more compact > * Common helper functions used to create the subdirectory contents > * Organization of the entries match the libsensors "related" field. > * Different parameters (min vs current) could have different scaling > * Filename wildcards and entry sorting are better. > > I think floating point is the best solution to the fractional values > that lm_sensors deals in. These values aren't integers. And we can try > and standardize on milli-volts and milli-degC, but for many sensors, > that's unnecessary and wastes range, and what about the values that > *are* integers. How do you distinguish between them and these scaled > values? The decimal point solves this and releases the requirement that > all values be expressed in milli-somethings. > > Further, if we restrict ourselves to scaling magnitudes that are greater > than or equal to 1, I think the conversions in and out of lm_sensors to > sysfs could be made simpler. But much of the code in i2c_*real is > dealing with the multiple values that can be found. With sysfs, we're > only allowed one value so these conversion functions would be much simpler. > > :v) >
