Re: libXO-ification - Why - and is it a symptom of deeper issues?

On 15 Nov, Allan Jude wrote:
> On 2015-11-15 13:06, Garrett Cooper wrote:
>> 
>>> On Nov 15, 2015, at 09:51, Andrey Chernov <ache_at_freebsd.org> wrote:
>>>
>>>> On 15.11.2015 20:37, Adrian Chadd wrote:
>>>>> On 15 November 2015 at 09:10, Dan Partelly <dan_partelly_at_rdsor.ro> wrote:
>>>>> Meaning, is that simple to push things in head , if somone does the work, even with with no proper review of the problem at hand , and the proposed solutions ?
>>>>
>>>> Nope and yup. The juniper folk had a solution to a problem multiple
>>>> people had requested work on, and their proposal was by far the
>>>> furthest along code and use wise.
>>>>
>>>> It's all fine and good making technical decisions based on drawings
>>>> and handwaving and philosophizing, but at some point someone has to do
>>>> the code. Juniper's libxo was the furthest along in implementation and
>>>> production.
>>>
>>> It seems it is the only and final argument for libXO existence. I
>>> remember 2 or 3 discussions against libXO spontaneously happens in the
>>> FreeBSD lists, all ended with that, approximately: "we already have the
>>> code and you have just speculations". Alternative and more architecture
>>> clean ideas, like making standalone template-oriented parser probably
>>> based on liXO, are never seriously considered, because nobody will code
>>> it, not for other reasons.
>> 
>> We lack a [dtd/json] spec for tools, so programming for xo'ification doesn't seems like the best idea in the world to me from a end-user sysadmin/developer perspective.
>> 
>> I could just as easily use standard tools like awk, grep, sed, and more advanced languages like perl or Python to parse output, and assuming output doesn't get a major rewrite, I'd just go with that method that's worked pretty well for me over the last 10 years of my career..
>> 
>> Cheers,
>> -NGie
>> _______________________________________________
>> freebsd-current_at_freebsd.org mailing list
>> https://lists.freebsd.org/mailman/listinfo/freebsd-current
>> To unsubscribe, send any mail to "freebsd-current-unsubscribe_at_freebsd.org"
>> 
> 
> The big difference is, a json parser isn't going to blow up if a new
> field gets added in the middle, and your awk/grep/sed script probably will.

Or more likely, if some value gets large causing adjacent columns run
together.  Or even if you the utility is used to display something where
some of the data contains whitespace.

How is an awk script going to cope with this:

# grep spacey /etc/passwd
spacey user:*:6666:6666:Update Builder:/tmp/spacey:/bin/csh
# su 'spacey user'
% cd ~
% touch 'a file'
% ls -l
total 1
-rw-r--r--  1 spacey user  6666  0 Nov 15 15:24 a file

vs a json parser:

% ls --libxo json -l
{"__version": "1", "file-information": {"directory": [{"total-blocks":1, "entry": [{"name":"a file","mode":"-rw-r--r--","mode_octal":644,"links":1,"user":"spacey user","group":"6666","type":"regular","size":0,"modify-time":1447629885}]}]}
}