On Tue, Feb 11, 2014 at 9:47 AM, Lee Jones <lee.jones@xxxxxxxxxx> wrote: >> >> >> On Mon, Feb 10, 2014 at 4:38 PM, Jean Delvare <jdelvare@xxxxxxx> wrote: >> >> >> > Additionally, dashes are explicitly forbidden in hwmon >> >> >> > device names. >> >> >> >> >> >> Also, where is that documented? >> >> > >> >> > In Documentation/hwmon/sysfs-interface: >> >> > >> >> > ********************* >> >> > * Global attributes * >> >> > ********************* >> >> > >> >> > name The chip name. >> >> > This should be a short, lowercase string, not containing >> >> > spaces nor dashes, representing the chip name. This is >> >> > the only mandatory attribute. >> >> > I2C devices get this attribute created automatically. >> >> > RO >> >> >> >> Time to revisit this decision.... >> >> >> >> So, based on the fact that children device names usually contain >> >> dashes, I do not understand why hwmon would be any special in this >> >> regard. It is possible that the hwmon developers have not faced much >> >> MFD situation before, and so, this was not considered to be handled >> >> like in other subsystems. >> >> >> >> I am proposing to change this "rule"... Any objection? >> > >> > Prior to proposing such an invasive change which is highly likely to >> > come up against heavy opposition, >> >> It is possible that someone does not understand why you think it may >> be invasive, right? Could you please explain the reason for that? > > The reason is a good one. In the kernel we make every attempt not to > break userspace. By that I mean _any_ userspace application. Userspace > applications which interface with the kernel can do so via a variety of > methods. One of those is Sysfs, where this name you are attempting to > change appears. The userspace applications already mentioned parse for > these devices, regex:ing for '-' separators. If you add an additional > '-' separator there is a chance that these applications will get > confused and break without warning. Only for new devices that they have not supported before, right? Provided, we apply the logic only for new drivers. Is it unacceptable to have an interim "obsolete" period for a while and the recommendation is the dash for the long future? Also, how about applications that would expect hwmon to behave the same way like all the rest? They also break with the current hwmon schema, right? > > -- > Lee Jones > Linaro STMicroelectronics Landing Team Lead > Linaro.org │ Open source software for ARM SoCs > Follow Linaro: Facebook | Twitter | Blog _______________________________________________ lm-sensors mailing list lm-sensors@xxxxxxxxxxxxxx http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
