Hi Nikolay,

Thanks for lots of follow-up examples :)

On Thu, Feb 4, 2016 at 8:33 AM, Nikolay Aleksandrovich Pavlov
<[email protected]> wrote:
> Since that thread is over I would reply here (did not reply previously
> because I did not realize that json and channel features are
> connected)

BTW, I didn't understand why json and channel features are connected.
I only talked about JSON feature.

>
>> The question is what to do with items that can't be converted to JSON.
>> I have been in doubt, whether to silently drop them or give an error.
>>
>> Although a missing item in an array is not according to the JSON
>> standard, it is very useful in practice:
>>
>> [1,,,,8]
>>
>> The missing items would be "undefined" in JavaScript.  In Vim they are
>> v:none.  Proper JSON would be:
>>
>> [1,null,null,null,8]
>>
>> That's more than twice as long.  Gets worse when there are more missing
>> items, up to five times as long.
>
> 1. Arrays like `[1, null, null, null, 8]` are very rarely used when
> communicating. So this is almost never “twice as long”. Especially
> with the planned subfeatures of the channel feature.
> 2. `[1,,,,8]` with the current parser is `[1, v:none, v:none, v:none,
> 8]`. If people are testing for null values using `if {val} is v:null`
> the fact that it can be written like this is absolutely useless. If
> people are using `if {val}` this can be as well written as
> `[1,0,0,0,8]`.
> 3. I have no problems in counting three null values, but commas are
> harder to count and they are usually visually skipped because of
> having very low importance.
> 4. Computer has no problems with either variants, performance impact
> is negligible.
> 5. Handling v:none in VimL in case somebody cares code adds *far* more
> ticks then parsing `null` in C code.
> 6. It is easy to miss v:none in cases like
>
>         [
>             "1"
>             , "2"
>             , "3",
>             , "4"
>         ]

This is very likely probable case!
The following JSON string leaves v:none at the end
if it is parsed by current jsondecode().

[
    "1",
    "2",
    "3",
    "4",
]

> 7. In msgpack the whole array is six bytes. JSON is more then three
> times as long. Non-JSON you propose is still 1⅓ longer. For IPC it is
> better. For user YAML is more readable (especially when one needs
> multi-line strings), and almost always can be made less verbose then
> JSON.
> 8. In javascript `[1,,,,8]` is `[1,undefined,undefined,undefined,8]`,
> not `[1,null,null,null,8]`.
>
>>
>> I propose to allow this extension to JSON.  However, it should not be
>> created accidentally, only when intentionally using v:none as an item.
>
> If documentation states that channel accepts JSON, it should accept
> JSON and not something else. I am not much fond of idea of
> jsondecode() extensions, but do not create *yet another* standard in
> interprocess communications, this action is worse then creating yet
> another non-strict parser.
>
>>
>> So, we should probably give an error when using a function, instead of
>> producing JSON that's not according to the standard.
>
> 2016-02-03 16:28 GMT+03:00 tyru <[email protected]>:
>> Dear Bram and Vimmers,
>>
>> I have read the below thread.
>>
>> [vim] jsonencode() does not work correctly with function references (#579)
>> https://groups.google.com/forum/#!msg/vim_dev/9rf5_YmPn28/qZKB3rKvCAAJ
>>
>> But, I couldn't understand why Vim has had to add v:none variable.
>>
>>> Although a missing item in an array is not according to the JSON
>>> standard, it is very useful in practice:
>> I don't think so.
>> If v:none variable has been added because *only* it is useful,
>> it should not be.
>> I don't think It should be added even breaking the JSON standard.
>>
>> Because, v:none and v:null is very similar.
>> If a user mistake v:none for v:null,
>> an invalid JSON (for strict JSON parser) will be generated!
>
> I can also confirm that I need to constantly remind myself what
> exactly I need to use.
>
>>
>>     jsonencode([1,v:none,v:none,4])
>>
>> "undefined" and "null" in JavaScript is totally a bad part.
>> Please don't follow that.
>
> It would be better if you have shown (pointed to an article) what is
> so bad here.

"undefined" and "null" are totally different things.
So "undefined === null" returns true, of course.
But both values mean often an absense of a value.
They are very confusing.

>
>>
>> Best wishes,
>>
>> tyru
>>
>> --
>> --
>> You received this message from the "vim_dev" maillist.
>> Do not top-post! Type your reply below the text you are replying to.
>> For more information, visit http://www.vim.org/maillist.php
>>
>> ---
>> You received this message because you are subscribed to the Google Groups 
>> "vim_dev" group.
>> To unsubscribe from this group and stop receiving emails from it, send an 
>> email to [email protected].
>> For more options, visit https://groups.google.com/d/optout.
>
> --
> --
> You received this message from the "vim_dev" maillist.
> Do not top-post! Type your reply below the text you are replying to.
> For more information, visit http://www.vim.org/maillist.php
>
> ---
> You received this message because you are subscribed to the Google Groups 
> "vim_dev" group.
> To unsubscribe from this group and stop receiving emails from it, send an 
> email to [email protected].
> For more options, visit https://groups.google.com/d/optout.

-- 
-- 
You received this message from the "vim_dev" maillist.
Do not top-post! Type your reply below the text you are replying to.
For more information, visit http://www.vim.org/maillist.php

--- 
You received this message because you are subscribed to the Google Groups 
"vim_dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Raspunde prin e-mail lui