On 11 Feb 2020 at 12:09p, Spectre pondered and said...
read () { TEMP="$(./temp | cut -d ' ' -f10)"
the script ./temp queries the thermometer and returns a string full of garbage aside from the temp. Failures show up as "firmware", "", or "v3.3:"
Oh. So that's meant to be _temperature_, not a generic name
for a _temporary_ variable. Got it now.
Note here that you _assign_ $TEMP to the string "firmware" if your conditions match. So if `$TEMP` was "v3.3:" it becomes "firmware".
Whats happening here is... firmware, v3.3: and "" are all failures, but I'm looping on firmware, so if it equals one of the other two, I change
it "firmware" A successful result looks like 21.5c. This logic mostly works, at least it does for "" and firmware, I don't comprende why v3.3: is an exception.
Ah, so you want to loop while what the earlier script read is NOT
a temperature. Ok, tracking.
As noted above, its only checking one status on the loop, so other failures need to be made into a logical failure. It does terminate
quite nicely in normal use. To be honest, when I started this, firmware was the only issue I was looking for, then I discovered the possible results.
Another approach to take is to probe for a _successful_ result,
and loop on anything else.
Thats partially right, there are 3 failures, but success is something
that doesn't match any known failure, so while it can look endless, it does exit on anything other than those values.
I see. I'd rewrite it to terminate on successfully finding
something that matches the temperature; the rest you simply
loop.
So long as the values mentioned above aren't met this does get executed, the file local.tmp is only the current time, temperature so overwriting
is a good thing. Its not a log as such. I have that being done
elsewhere.
I might have to cut it out, and pore over it a bit more in a shell, but I'm not entirely with your rewrite.. :/
Fair enough. Understanding the problem a little more, it
sounds like you want to:
1. Try and read the current temperature.
2. If whatever you read doesn't look like a temperature, sleep
for a bit and try again.
3. Otherwise, echo the temperature read and the time to a file.
The simplest way to do this is probably to pipe the cut output
from your `temp` script through `grep` to match something that
looks like a temperature. Here's my updated rewrite:
: fat-dragon; rm .??*
: fat-dragon; ls -A
scr
temp
: fat-dragon; cat temp
#!/bin/sh
if ! [ -f .blarg ]; then
echo ' v3.3:'
touch .blarg
elif ! [ -f .blat ]; then
echo
touch .blat
elif ! [ -f .blah ]; then
echo ' firmware'
touch .blah
else
echo 25.2C
fi
: fat-dragon; cat scr
#!/bin/sh
temperature() {
./temp | cut -d ' ' -f 10 | egrep '^-?[0-9]+\.[0-9]+[Cc]$'
}
now() {
date +%R
}
temp=$(temperature)
while [ -z $temp ]; do
sleep 5
temp=$(temperature)
done
echo "Frankston $temp at $(now)"
: fat-dragon; date; ./scr; date
Tue Feb 11 02:46:27 UTC 2020
Frankston 25.2C at 02:46
Tue Feb 11 02:46:42 UTC 2020
: fat-dragon; ls -A
.blah
.blarg
.blat
scr
temp
: fat-dragon; rm .??*
: fat-dragon;
Here, the magic is in the `egrep`: the point here is to
use a (POSIX extended) regular expression to find something
that "looks" like a temperature, and permit only those
strings from the updated `temperature` function. If we
see anything that doesn't look like a temperature, then
the output from `temperature` will be empty, and we can
test for that using the `-z` argument to `test` (`[` is
another word for `test`). We'll loop as long as the
temperature we read is empty; that is, until we get a
useful value from the `temp` script.
--- Mystic BBS v1.12 A44 2020/02/04 (Windows/32)
* Origin: Agency BBS | Dunedin, New Zealand | agency.bbs.nz (21:1/101)