On 20/01/2021 17:19, Martin Gregorie wrote:
> On Wed, 20 Jan 2021 15:49:49 +0000, Pancho wrote:
>
>> On 20/01/2021 13:02, Martin Gregorie wrote:
>>
>>> So, questions for Pancho:
>>>
>>> - how did you create your input files?
>>>
>>> - Is there any chance that they don't contain any 'newline' characters?
>>>
>>> I'm asking because its quite possible that awk would stall, waiting for
>>> a the next character, if it had read several KB of data without finding
>>> a newline character or EOF.
>>>
>>>
>>>
>> I used:
>>
>> tail -f testfile | mawk...
>>
>> I used vi to generate files, with newlines, which I appended into the
>> testfile, or just echo "stuff..." appended into the test file.
>>
>> But Richard's example is perfect to demonstrate the problem:
>>
>>   >> $ (seq 9999 | head -c 4095; sleep 2; echo) | mawk '{print}'
>>
>> Not only does seq quickly generate a byte stream, but it allows you to
>> see exactly what byte you are on. I'm not sure why he used the brackets
>> () but I left them in as they don't hurt.
>>
>> I guess he used the echo because he is habitually tidy :-).
>>
>> Maybe the answer is that for reliability I should have used perl instead
>> of awk?
>>
> Dunno about that: awk takes a bit of getting used to because its a
> specialised tool for extracting data from text files: I suggested using
> it in this case because its brilliant for tasks like scanning log files
> for a set of error messages and doing appropriate stuff for each one it
> spots. As you've seen, the code needed to recognise 'battery low'
> warnings in, say, /var/log/messages and issue a 'shutdown -h now' command
> if one is found would be a trivially small program because it
> automatically reads lines and splits them into words before using a regex
> to trigger actions on any lines that the regex matches: about all you
> need to write is the regexes and the code that each of them triggers.
>
> By comparison C, Java, Perl or Python, ... are all general purpose
> programming languages and so you need to code the file reading loop and
> (probably) the scan routine that recognises interesting lines in the file
> as well as the code to do something useful when a line is recognised.
>
> IOW, to spot a 'battery low' message in the /var/log/messages logfile and
> issue a STOP command the Raspbian would be a one liner in awk:
>
> awk -- '/battery low/ { system("shutdown -h now") }' /var/log/messages
>
> as compared with at least 10-20 lines in any of the other languages I
> mentioned - and they'll take longer to write simply because the loop and
> the line parser will need to be coded and debugged.
>

tail -f log | perl -ne 'system('shutdown','-h now') if /battery low/'


> The above, of course, is assuming that you don't want to simply build the
> shutdown command into the program that's watching battery voltage
>
> A lot of programmers would do it that way while others, of which I'm one,
> like to keep different activities in single purpose programs - and,
> though its unlikely in this particular case, would keep them separate
> because you might someday need to let another program trigger the boojum
> as well as the one you're designing right now.
>


I would prefer an explicit observer mechanism, bundled with journalctl,
but clearly I'm a MS fanboi.

It's not the behaviour of mawk I object to, per se, but the fact that
awk can change behaviour so easily with system config. Really speaking,
I've only been willing to touch Linux in the last 5-6 years (after a 20
year Unix hiatus) because Vagrant and then Docker allowed me to create a
reliable environment.

--- SoupGate-Win32 v1.05