Alistair John Strachan wrote: > On Thursday 08 January 2009 18:09:40 Alistair John Strachan wrote: >> On Thursday 08 January 2009 17:09:43 Jean Delvare wrote: > [snip] >>> Might be worth reading this discussion thread again: >>> http://lists.lm-sensors.org/pipermail/lm-sensors/2008-October/024522.html >>> >>> We gave up on fixing it back then, but apparently there really is a >>> need. Alistair, do you have a patch? > > Find attached an alternative method which solves (I believe only potential) > problems Jean pointed out at the time of the last thread. I think it's safe. > > I've compromised by finding the first open bracket (it's fine if there are no > brackets) and trimming any space from the sub-string. For example, here are > the existing transformations: > > "AW9D-MAX (Intel i975-ICH7)" -> "AW9D-MAX" > "AT8 32X(ATI RD580-ULI M1575)" -> "AT8 32X" > "IP35 Pro(Intel P35-ICH9R)" -> "IP35 Pro" > "IN9 32X MAX(680i-MCP55PXE)" -> "IN9 32X MAX" (adding RSN) > > And here are some hypothetical ones: > > "AW9D (Blah)" -> "AW9D" > "Abit Magic" -> "Abit Magic" > > In the AW9D case, "AW9D" and "AW9D-MAX" are not considered to be the same > motherboard, because the sub-string's length must exactly match that of the > string in the motherboard entry. This is different to regular "strncasecmp" > where they would be considered the same. > > Find the patch attached. I'll probably go for this version if nobody finds any > issues with it, otherwise the original patch I sent out is sufficient (since > at the moment there are no boards falling into this hypothetical trap). Good catch Jean! The new version looks good to me. Regards, Hans
