Re: [ANN] Penlight 1.12.0 released

[Scott Morgan <blumf@...>]

On 10/01/2022 15:18, steve donovan wrote:
> On Mon, Jan 10, 2022 at 5:02 PM Scott Morgan <blumf@blueyonder.co.uk> wrote:
>>> libraries should be used for parsing XML, and not a general propose
>>> utility library like Penlight, hence the decision.
>> This is a shame, as most the XML libs out there seem to be abandoned.
> 
> I have personally veered all over the map on this question - XML is
> both big and (increasingly) legacy. There are informal uses of XML -
> let us call it 'Pointy-Bracket Data Format' and the
> full-on-plus-schema toolchain. Obviously PL should never be used for
> 'Big XML', I suppose I was always a skeptic who appreciated the simple
> use of the format.

Yeah, the 'Big XML' stuff is ridiculous, but the straight forward data
medium is very useful as XML parsers/writers are around most places
(browsers, .Net, Java, many DB systems) so interoperability is good.

> The question about hese libraries is this: perhaps they are finished?

That may be part of it; unless bugs are found, there's probably not much
to do.

> PS and sure enough, PL never did JSON although that's the common
> interchange format these days (for better or worse). I was always a
> fan of LTON (Lua Table Object Notation) which was discussed some years
> ago. As long as you don't just use load()! I don't know of any fast
> implementations.

Never been convinced by JSON (and again, not much in the Lua ecosystem,
when I looked), and there are a whole bunch of other options (Google
released a nice looking and efficient binary format a while back[1]),
but you end up with XKCD #927.

Personally, interoperability is the main need, and XML wins out there,
with JSON coming a close second.

Scott

[1] https://developers.google.com/protocol-buffers