* [lm-sensors] "Built-in" via686a sensors names
@ 2006-10-30 6:52 Ian Pilcher
2006-10-31 8:19 ` Jean Delvare
` (2 more replies)
0 siblings, 3 replies; 4+ messages in thread
From: Ian Pilcher @ 2006-10-30 6:52 UTC (permalink / raw)
To: lm-sensors
The gkrellm package in Fedora Extras 6 has been patched to use
libsensors. This causes it to fail to detect the voltage sensors on my
Abit VP6 motherboard:
[pilcher at home ~]$ gkrellm -d 0x80
--- GKrellM 2.2.9 ---
hddtemp_daemon_read: can't connect to 127.0.0.1:7634.
libsensors error determining type for: via686a at 6000/2.0V
libsensors error determining type for: via686a at 6000/2.5V
libsensors error determining type for: via686a at 6000/3.3V
libsensors error determining type for: via686a at 6000/5.0V
libsensors error determining type for: via686a at 6000/12V
libsensors error determining type for: via686a at 6000/alarms
gkrellm expects voltage sensors to to be named "inX", but it is getting
back the "built-in" names such as "2.0V", etc. (Built-in names which,
BTW, are incorrect for my motherboard.)
While I understand the desire to provide meaningful output, this seems
to be a case where the library is being too clever by half. Shouldn't
the library just return the "raw" name (in0, etc.) and let the human-
readable names be defined in sensors.conf?
Or am I missing something...
Thanks!
--
====================================
Ian Pilcher i.pilcher at comcast.net
====================================
^ permalink raw reply [flat|nested] 4+ messages in thread
* [lm-sensors] "Built-in" via686a sensors names
2006-10-30 6:52 [lm-sensors] "Built-in" via686a sensors names Ian Pilcher
@ 2006-10-31 8:19 ` Jean Delvare
2006-10-31 18:29 ` Ian Pilcher
2006-10-31 21:15 ` Jean Delvare
2 siblings, 0 replies; 4+ messages in thread
From: Jean Delvare @ 2006-10-31 8:19 UTC (permalink / raw)
To: lm-sensors
Hi Ian,
On 10/30/2006, Ian Pilcher wrote:
>The gkrellm package in Fedora Extras 6 has been patched to use
>libsensors. This causes it to fail to detect the voltage sensors on my
>Abit VP6 motherboard:
>
>[pilcher at home ~]$ gkrellm -d 0x80
>--- GKrellM 2.2.9 ---
>hddtemp_daemon_read: can't connect to 127.0.0.1:7634.
>libsensors error determining type for: via686a at 6000/2.0V
>libsensors error determining type for: via686a at 6000/2.5V
>libsensors error determining type for: via686a at 6000/3.3V
>libsensors error determining type for: via686a at 6000/5.0V
>libsensors error determining type for: via686a at 6000/12V
>libsensors error determining type for: via686a at 6000/alarms
>
>gkrellm expects voltage sensors to to be named "inX", but it is getting
>back the "built-in" names such as "2.0V", etc. (Built-in names which,
>BTW, are incorrect for my motherboard.)
>
>While I understand the desire to provide meaningful output, this seems
>to be a case where the library is being too clever by half. Shouldn't
>the library just return the "raw" name (in0, etc.) and let the human-
>readable names be defined in sensors.conf?
libsensors returns the old 2.4 kernel symbol names, which were not
standard. Programs should not make any assumption about these symbol
names, they are arbitrary.
This was admittedly a design error, which is why the names where
standardized in kernel 2.6, but for now applications still go through
the old names in libsensors for compatibility reasons (sensors.conf
references the old names). We hope to have an alternative access method
in lm_sensors 2.11.0, which would solve the problem, but it will only
support 2.6 kernels, and will require a new sensors.conf file.
>Or am I missing something...
This is a bug in Fedora and/or gkrellm. What do you expect from us by
reporting it on the lm_sensors list?
--
Jean Delvare
^ permalink raw reply [flat|nested] 4+ messages in thread
* [lm-sensors] "Built-in" via686a sensors names
2006-10-30 6:52 [lm-sensors] "Built-in" via686a sensors names Ian Pilcher
2006-10-31 8:19 ` Jean Delvare
@ 2006-10-31 18:29 ` Ian Pilcher
2006-10-31 21:15 ` Jean Delvare
2 siblings, 0 replies; 4+ messages in thread
From: Ian Pilcher @ 2006-10-31 18:29 UTC (permalink / raw)
To: lm-sensors
Jean Delvare wrote:
> This is a bug in Fedora and/or gkrellm. What do you expect from us by
> reporting it on the lm_sensors list?
I guess it would be a bug in gkrellm, since it's making the assumption
that a voltage sensor will be named "inX". OTOH, I'm not really sure
what else an application is supposed to do, since there doesn't appear
to be an API to determine the type of a sensor.
Mostly, I wanted to start a discussion about whether the current
behavior is useful. From what you say about 2.11.0, it sounds like the
problems it causes have been recognized, which is good news.
Thanks!
--
====================================
Ian Pilcher i.pilcher at comcast.net
====================================
^ permalink raw reply [flat|nested] 4+ messages in thread
* [lm-sensors] "Built-in" via686a sensors names
2006-10-30 6:52 [lm-sensors] "Built-in" via686a sensors names Ian Pilcher
2006-10-31 8:19 ` Jean Delvare
2006-10-31 18:29 ` Ian Pilcher
@ 2006-10-31 21:15 ` Jean Delvare
2 siblings, 0 replies; 4+ messages in thread
From: Jean Delvare @ 2006-10-31 21:15 UTC (permalink / raw)
To: lm-sensors
Hi Ian,
On 10/31/2006, Ian Pilcher wrote:
> Jean Delvare wrote:
> > This is a bug in Fedora and/or gkrellm. What do you expect from us by
> > reporting it on the lm_sensors list?
>
> I guess it would be a bug in gkrellm, since it's making the assumption
> that a voltage sensor will be named "inX". OTOH, I'm not really sure
> what else an application is supposed to do, since there doesn't appear
> to be an API to determine the type of a sensor.
Indeed, this part is missing from the current API, has been identified as
such, and should be added in a near future. Right now, applications must
either include device-specific code to format the values properly (as
sensors does), or display only numbers without units (as sensors -u
does). I guess this is the major reason why the author of gkrellm never
wanted to use libsensors in the first place.
> Mostly, I wanted to start a discussion about whether the current
> behavior is useful. From what you say about 2.11.0, it sounds like the
> problems it causes have been recognized, which is good news.
The current behavior must be somewhat useful, as a dozen applications
have been using it for many years. But it's admittedly suboptimal.
Something better should be available soon (say mid-2007), making it
possible for application authors to write only generic code and still
handle all devices propely with nicely formatted output (Linux 2.6 only.)
--
Jean Delvare
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2006-10-31 21:15 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2006-10-30 6:52 [lm-sensors] "Built-in" via686a sensors names Ian Pilcher
2006-10-31 8:19 ` Jean Delvare
2006-10-31 18:29 ` Ian Pilcher
2006-10-31 21:15 ` Jean Delvare
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.